AI is developing all the time. Here are some picks from several articles what is expected to happen in AI and around it in 2025. Here are picks from various articles, the texts are picks from the article edited and in some cases translated for clarity.
AI in 2025: Five Defining Themes
https://news.sap.com/2025/01/ai-in-2025-defining-themes/
Artificial intelligence (AI) is accelerating at an astonishing pace, quickly moving from emerging technologies to impacting how businesses run. From building AI agents to interacting with technology in ways that feel more like a natural conversation, AI technologies are poised to transform how we work.
But what exactly lies ahead?
1. Agentic AI: Goodbye Agent Washing, Welcome Multi-Agent Systems
AI agents are currently in their infancy. While many software vendors are releasing and labeling the first “AI agents” based on simple conversational document search, advanced AI agents that will be able to plan, reason, use tools, collaborate with humans and other agents, and iteratively reflect on progress until they achieve their objective are on the horizon. The year 2025 will see them rapidly evolve and act more autonomously. More specifically, 2025 will see AI agents deployed more readily “under the hood,” driving complex agentic workflows.
In short, AI will handle mundane, high-volume tasks while the value of human judgement, creativity, and quality outcomes will increase.
2. Models: No Context, No Value
Large language models (LLMs) will continue to become a commodity for vanilla generative AI tasks, a trend that has already started. LLMs are drawing on an increasingly tapped pool of public data scraped from the internet. This will only worsen, and companies must learn to adapt their models to unique, content-rich data sources.
We will also see a greater variety of foundation models that fulfill different purposes. Take, for example, physics-informed neural networks (PINNs), which generate outcomes based on predictions grounded in physical reality or robotics. PINNs are set to gain more importance in the job market because they will enable autonomous robots to navigate and execute tasks in the real world.
Models will increasingly become more multimodal, meaning an AI system can process information from various input types.
3. Adoption: From Buzz to Business
While 2024 was all about introducing AI use cases and their value for organizations and individuals alike, 2025 will see the industry’s unprecedented adoption of AI specifically for businesses. More people will understand when and how to use AI, and the technology will mature to the point where it can deal with critical business issues such as managing multi-national complexities. Many companies will also gain practical experience working for the first time through issues like AI-specific legal and data privacy terms (compared to when companies started moving to the cloud 10 years ago), building the foundation for applying the technology to business processes.
4. User Experience: AI Is Becoming the New UI
AI’s next frontier is seamlessly unifying people, data, and processes to amplify business outcomes. In 2025, we will see increased adoption of AI across the workforce as people discover the benefits of humans plus AI.
This means disrupting the classical user experience from system-led interactions to intent-based, people-led conversations with AI acting in the background. AI copilots will become the new UI for engaging with a system, making software more accessible and easier for people. AI won’t be limited to one app; it might even replace them one day. With AI, frontend, backend, browser, and apps are blurring. This is like giving your AI “arms, legs, and eyes.”
5. Regulation: Innovate, Then Regulate
It’s fair to say that governments worldwide are struggling to keep pace with the rapid advancements in AI technology and to develop meaningful regulatory frameworks that set appropriate guardrails for AI without compromising innovation.
12 AI predictions for 2025
This year we’ve seen AI move from pilots into production use cases. In 2025, they’ll expand into fully-scaled, enterprise-wide deployments.
https://www.cio.com/article/3630070/12-ai-predictions-for-2025.html
This year we’ve seen AI move from pilots into production use cases. In 2025, they’ll expand into fully-scaled, enterprise-wide deployments.
1. Small language models and edge computing
Most of the attention this year and last has been on the big language models — specifically on ChatGPT in its various permutations, as well as competitors like Anthropic’s Claude and Meta’s Llama models. But for many business use cases, LLMs are overkill and are too expensive, and too slow, for practical use.
“Looking ahead to 2025, I expect small language models, specifically custom models, to become a more common solution for many businesses,”
2. AI will approach human reasoning ability
In mid-September, OpenAI released a new series of models that thinks through problems much like a person would, it claims. The company says it can achieve PhD-level performance in challenging benchmark tests in physics, chemistry, and biology. For example, the previous best model, GPT-4o, could only solve 13% of the problems on the International Mathematics Olympiad, while the new reasoning model solved 83%.
If AI can reason better, then it will make it possible for AI agents to understand our intent, translate that into a series of steps, and do things on our behalf, says Gartner analyst Arun Chandrasekaran. “Reasoning also helps us use AI as more of a decision support system,”
3. Massive growth in proven use cases
This year, we’ve seen some use cases proven to have ROI, says Monteiro. In 2025, those use cases will see massive adoption, especially if the AI technology is integrated into the software platforms that companies are already using, making it very simple to adopt.
“The fields of customer service, marketing, and customer development are going to see massive adoption,”
4. The evolution of agile development
The agile manifesto was released in 2001 and, since then, the development philosophy has steadily gained over the previous waterfall style of software development.
“For the last 15 years or so, it’s been the de-facto standard for how modern software development works,”
5. Increased regulation
At the end of September, California governor Gavin Newsom signed a law requiring gen AI developers to disclose the data they used to train their systems, which applies to developers who make gen AI systems publicly available to Californians. Developers must comply by the start of 2026.
There are also regulations about the use of deep fakes, facial recognition, and more. The most comprehensive law, the EU’s AI Act, which went into effect last summer, is also something that companies will have to comply with starting in mid-2026, so, again, 2025 is the year when they will need to get ready.
6. AI will become accessible and ubiquitous
With gen AI, people are still at the stage of trying to figure out what gen AI is, how it works, and how to use it.
“There’s going to be a lot less of that,” he says. But gen AI will become ubiquitous and seamlessly woven into workflows, the way the internet is today.
7. Agents will begin replacing services
Software has evolved from big, monolithic systems running on mainframes, to desktop apps, to distributed, service-based architectures, web applications, and mobile apps. Now, it will evolve again, says Malhotra. “Agents are the next phase,” he says. Agents can be more loosely coupled than services, making these architectures more flexible, resilient and smart. And that will bring with it a completely new stack of tools and development processes.
8. The rise of agentic assistants
In addition to agents replacing software components, we’ll also see the rise of agentic assistants, adds Malhotra. Take for example that task of keeping up with regulations.
Today, consultants get continuing education to stay abreast of new laws, or reach out to colleagues who are already experts in them. It takes time for the new knowledge to disseminate and be fully absorbed by employees.
“But an AI agent can be instantly updated to ensure that all our work is compliant with the new laws,” says Malhotra. “This isn’t science fiction.”
9. Multi-agent systems
Sure, AI agents are interesting. But things are going to get really interesting when agents start talking to each other, says Babak Hodjat, CTO of AI at Cognizant. It won’t happen overnight, of course, and companies will need to be careful that these agentic systems don’t go off the rails.
Companies such as Sailes and Salesforce are already developing multi-agent workflows.
10. Multi-modal AI
Humans and the companies we build are multi-modal. We read and write text, we speak and listen, we see and we draw. And we do all these things through time, so we understand that some things come before other things. Today’s AI models are, for the most part, fragmentary. One can create images, another can only handle text, and some recent ones can understand or produce video.
11. Multi-model routing
Not to be confused with multi-modal AI, multi-modal routing is when companies use more than one LLM to power their gen AI applications. Different AI models are better at different things, and some are cheaper than others, or have lower latency. And then there’s the matter of having all your eggs in one basket.
“A number of CIOs I’ve spoken with recently are thinking about the old ERP days of vendor lock,” says Brett Barton, global AI practice leader at Unisys. “And it’s top of mind for many as they look at their application portfolio, specifically as it relates to cloud and AI capabilities.”
Diversifying away from using just a single model for all use cases means a company is less dependent on any one provider and can be more flexible as circumstances change.
12. Mass customization of enterprise software
Today, only the largest companies, with the deepest pockets, get to have custom software developed specifically for them. It’s just not economically feasible to build large systems for small use cases.
“Right now, people are all using the same version of Teams or Slack or what have you,” says Ernst & Young’s Malhotra. “Microsoft can’t make a custom version just for me.” But once AI begins to accelerate the speed of software development while reducing costs, it starts to become much more feasible.
9 IT resolutions for 2025
https://www.cio.com/article/3629833/9-it-resolutions-for-2025.html
1. Innovate
“We’re embracing innovation,”
2. Double down on harnessing the power of AI
Not surprisingly, getting more out of AI is top of mind for many CIOs.
“I am excited about the potential of generative AI, particularly in the security space,”
3. And ensure effective and secure AI rollouts
“AI is everywhere, and while its benefits are extensive, implementing it effectively across a corporation presents challenges. Balancing the rollout with proper training, adoption, and careful measurement of costs and benefits is essential, particularly while securing company assets in tandem,”
4. Focus on responsible AI
The possibilities of AI grow by the day — but so do the risks.
“My resolution is to mature in our execution of responsible AI,”
“AI is the new gold and in order to truly maximize it’s potential, we must first have the proper guardrails in place. Taking a human-first approach to AI will help ensure our state can maintain ethics while taking advantage of the new AI innovations.”
5. Deliver value from generative AI
As organizations move from experimenting and testing generative AI use cases, they’re looking for gen AI to deliver real business value.
“As we go into 2025, we’ll continue to see the evolution of gen AI. But it’s no longer about just standing it up. It’s more about optimizing and maximizing the value we’re getting out of gen AI,”
6. Empower global talent
Although harnessing AI is a top objective for Morgan Stanley’s Wetmur, she says she’s equally committed to harnessing the power of people.
7. Create a wholistic learning culture
Wetmur has another talent-related objective: to create a learning culture — not just in her own department but across all divisions.
8. Deliver better digital experiences
Deltek’s Cilsick has her sights set on improving her company’s digital employee experience, believing that a better DEX will yield benefits in multiple ways.
Cilsick says she first wants to bring in new technologies and automation to “make things as easy as possible,” mirroring the digital experiences most workers have when using consumer technologies.
“It’s really about leveraging tech to make sure [employees] are more efficient and productive,”
“In 2025 my primary focus as CIO will be on transforming operational efficiency, maximizing business productivity, and enhancing employee experiences,”
9. Position the company for long-term success
Lieberman wants to look beyond 2025, saying another resolution for the year is “to develop a longer-term view of our technology roadmap so that we can strategically decide where to invest our resources.”
“My resolutions for 2025 reflect the evolving needs of our organization, the opportunities presented by AI and emerging technologies, and the necessity to balance innovation with operational efficiency,”
Lieberman aims to develop AI capabilities to automate routine tasks.
“Bots will handle common inquiries ranging from sales account summaries to HR benefits, reducing response times and freeing up resources for strategic initiatives,”
Not just hype — here are real-world use cases for AI agents
https://venturebeat.com/ai/not-just-hype-here-are-real-world-use-cases-for-ai-agents/
Just seven or eight months ago, when a customer called in to or emailed Baca Systems with a service question, a human agent handling the query would begin searching for similar cases in the system and analyzing technical documents.
This process would take roughly five to seven minutes; then the agent could offer the “first meaningful response” and finally begin troubleshooting.
But now, with AI agents powered by Salesforce, that time has been shortened to as few as five to 10 seconds.
Now, instead of having to sift through databases for previous customer calls and similar cases, human reps can ask the AI agent to find the relevant information. The AI runs in the background and allows humans to respond right away, Russo noted.
AI can serve as a sales development representative (SDR) to send out general inquires and emails, have a back-and-forth dialogue, then pass the prospect to a member of the sales team, Russo explained.
But once the company implements Salesforce’s Agentforce, a customer needing to modify an order will be able to communicate their needs with AI in natural language, and the AI agent will automatically make adjustments. When more complex issues come up — such as a reconfiguration of an order or an all-out venue change — the AI agent will quickly push the matter up to a human rep.
Open Source in 2025: Strap In, Disruption Straight Ahead
Look for new tensions to arise in the New Year over licensing, the open source AI definition, security and compliance, and how to pay volunteer maintainers.
https://thenewstack.io/open-source-in-2025-strap-in-disruption-straight-ahead/
The trend of widely used open source software moving to more restrictive licensing isn’t new.
In addition to the demands of late-stage capitalism and impatient investors in companies built on open source tools, other outside factors are pressuring the open source world. There’s the promise/threat of generative AI, for instance. Or the shifting geopolitical landscape, which brings new security concerns and governance regulations.
What’s ahead for open source in 2025?
More Consolidation, More Licensing Changes
The Open Source AI Debate: Just Getting Started
Security and Compliance Concerns Will Rise
Paying Maintainers: More Cash, Creativity Needed
Kyberturvallisuuden ja tekoälyn tärkeimmät trendit 2025
https://www.uusiteknologia.fi/2024/11/20/kyberturvallisuuden-ja-tekoalyn-tarkeimmat-trendit-2025/
1. Cyber infrastructure will be centered on a single, unified security platform
2. Big data will give an edge against new entrants
3. AI’s integrated role in 2025 means building trust, governance engagement, and a new kind of leadership
4. Businesses will adopt secure enterprise browsers more widely
5. AI’s energy implications will be more widely recognized in 2025
6. Quantum realities will become clearer in 2025
7. Security and marketing leaders will work more closely together
Presentation: For 2025, ‘AI eats the world’.
https://www.ben-evans.com/presentations
Just like other technologies that have gone before, such as cloud and cybersecurity automation, right now AI lacks maturity.
https://www.securityweek.com/ai-implementing-the-right-technology-for-the-right-use-case/
If 2023 and 2024 were the years of exploration, hype and excitement around AI, 2025 (and 2026) will be the year(s) that organizations start to focus on specific use cases for the most productive implementations of AI and, more importantly, to understand how to implement guardrails and governance so that it is viewed as less of a risk by security teams and more of a benefit to the organization.
Businesses are developing applications that add Large Language Model (LLM) capabilities to provide superior functionality and advanced personalization
Employees are using third party GenAI tools for research and productivity purposes
Developers are leveraging AI-powered code assistants to code faster and meet challenging production deadlines
Companies are building their own LLMs for internal use cases and commercial purposes.
AI is still maturing
However, just like other technologies that have gone before, such as cloud and cybersecurity automation, right now AI lacks maturity. Right now, we very much see AI in this “peak of inflated expectations” phase and predict that it will dip into the “trough of disillusionment”, where organizations realize that it is not the silver bullet they thought it would be. In fact, there are already signs of cynicism as decision-makers are bombarded with marketing messages from vendors and struggle to discern what is a genuine use case and what is not relevant for their organization.
There is also regulation that will come into force, such as the EU AI Act, which is a comprehensive legal framework that sets out rules for the development and use of AI.
AI certainly won’t solve every problem, and it should be used like automation, as part of a collaborative mix of people, process and technology. You simply can’t replace human intuition with AI, and many new AI regulations stipulate that human oversight is maintained.
7 Splunk Predictions for 2025
https://www.splunk.com/en_us/form/future-predictions.html
AI: Projects must prove their worth to anxious boards or risk defunding, and LLMs will go small to reduce operating costs and environmental impact.
OpenAI, Google and Anthropic Are Struggling to Build More Advanced AI
Three of the leading artificial intelligence companies are seeing diminishing returns from their costly efforts to develop newer models.
https://www.bloomberg.com/news/articles/2024-11-13/openai-google-and-anthropic-are-struggling-to-build-more-advanced-ai
Sources: OpenAI, Google, and Anthropic are all seeing diminishing returns from costly efforts to build new AI models; a new Gemini model misses internal targets
It Costs So Much to Run ChatGPT That OpenAI Is Losing Money on $200 ChatGPT Pro Subscriptions
https://futurism.com/the-byte/openai-chatgpt-pro-subscription-losing-money?fbclid=IwY2xjawH8epVleHRuA2FlbQIxMQABHeggEpKe8ZQfjtPRC0f2pOI7A3z9LFtFon8lVG2VAbj178dkxSQbX_2CJQ_aem_N_ll3ETcuQ4OTRrShHqNGg
In a post on X-formerly-Twitter, CEO Sam Altman admitted an “insane” fact: that the company is “currently losing money” on ChatGPT Pro subscriptions, which run $200 per month and give users access to its suite of products including its o1 “reasoning” model.
“People use it much more than we expected,” the cofounder wrote, later adding in response to another user that he “personally chose the price and thought we would make some money.”
Though Altman didn’t explicitly say why OpenAI is losing money on these premium subscriptions, the issue almost certainly comes down to the enormous expense of running AI infrastructure: the massive and increasing amounts of electricity needed to power the facilities that power AI, not to mention the cost of building and maintaining those data centers. Nowadays, a single query on the company’s most advanced models can cost a staggering $1,000.
Tekoäly edellyttää yhä nopeampia verkkoja
https://etn.fi/index.php/opinion/16974-tekoaely-edellyttaeae-yhae-nopeampia-verkkoja
A resilient digital infrastructure is critical to effectively harnessing telecommunications networks for AI innovations and cloud-based services. The increasing demand for data-rich applications related to AI requires a telecommunications network that can handle large amounts of data with low latency, writes Carl Hansson, Partner Solutions Manager at Orange Business.
AI’s Slowdown Is Everyone Else’s Opportunity
Businesses will benefit from some much-needed breathing space to figure out how to deliver that all-important return on investment.
https://www.bloomberg.com/opinion/articles/2024-11-20/ai-slowdown-is-everyone-else-s-opportunity
Näin sirumarkkinoilla käy ensi vuonna
https://etn.fi/index.php/13-news/16984-naein-sirumarkkinoilla-kaey-ensi-vuonna
The growing demand for high-performance computing (HPC) for artificial intelligence and HPC computing continues to be strong, with the market set to grow by more than 15 percent in 2025, IDC estimates in its recent Worldwide Semiconductor Technology Supply Chain Intelligence report.
IDC predicts eight significant trends for the chip market by 2025.
1. AI growth accelerates
2. Asia-Pacific IC Design Heats Up
3. TSMC’s leadership position is strengthening
4. The expansion of advanced processes is accelerating.
5. Mature process market recovers
6. 2nm Technology Breakthrough
7. Restructuring the Packaging and Testing Market
8. Advanced packaging technologies on the rise
2024: The year when MCUs became AI-enabled
https://www-edn-com.translate.goog/2024-the-year-when-mcus-became-ai-enabled/?fbclid=IwZXh0bgNhZW0CMTEAAR1_fEakArfPtgGZfjd-NiPd_MLBiuHyp9qfiszczOENPGPg38wzl9KOLrQ_aem_rLmf2vF2kjDIFGWzRVZWKw&_x_tr_sl=en&_x_tr_tl=fi&_x_tr_hl=fi&_x_tr_pto=wapp
The AI party in the MCU space started in 2024, and in 2025, it is very likely that there will be more advancements in MCUs using lightweight AI models.
Adoption of AI acceleration features is a big step in the development of microcontrollers. The inclusion of AI features in microcontrollers started in 2024, and it is very likely that in 2025, their features and tools will develop further.
Just like other technologies that have gone before, such as cloud and cybersecurity automation, right now AI lacks maturity.
https://www.securityweek.com/ai-implementing-the-right-technology-for-the-right-use-case/
If 2023 and 2024 were the years of exploration, hype and excitement around AI, 2025 (and 2026) will be the year(s) that organizations start to focus on specific use cases for the most productive implementations of AI and, more importantly, to understand how to implement guardrails and governance so that it is viewed as less of a risk by security teams and more of a benefit to the organization.
Businesses are developing applications that add Large Language Model (LLM) capabilities to provide superior functionality and advanced personalization
Employees are using third party GenAI tools for research and productivity purposes
Developers are leveraging AI-powered code assistants to code faster and meet challenging production deadlines
Companies are building their own LLMs for internal use cases and commercial purposes.
AI is still maturing
AI Regulation Gets Serious in 2025 – Is Your Organization Ready?
While the challenges are significant, organizations have an opportunity to build scalable AI governance frameworks that ensure compliance while enabling responsible AI innovation.
https://www.securityweek.com/ai-regulation-gets-serious-in-2025-is-your-organization-ready/
Similar to the GDPR, the EU AI Act will take a phased approach to implementation. The first milestone arrives on February 2, 2025, when organizations operating in the EU must ensure that employees involved in AI use, deployment, or oversight possess adequate AI literacy. Thereafter from August 1 any new AI models based on GPAI standards must be fully compliant with the act. Also similar to GDPR is the threat of huge fines for non-compliance – EUR 35 million or 7 percent of worldwide annual turnover, whichever is higher.
While this requirement may appear manageable on the surface, many organizations are still in the early stages of defining and formalizing their AI usage policies.
Later phases of the EU AI Act, expected in late 2025 and into 2026, will introduce stricter requirements around prohibited and high-risk AI applications. For organizations, this will surface a significant governance challenge: maintaining visibility and control over AI assets.
Tracking the usage of standalone generative AI tools, such as ChatGPT or Claude, is relatively straightforward. However, the challenge intensifies when dealing with SaaS platforms that integrate AI functionalities on the backend. Analysts, including Gartner, refer to this as “embedded AI,” and its proliferation makes maintaining accurate AI asset inventories increasingly complex.
Where frameworks like the EU AI Act grow more complex is their focus on ‘high-risk’ use cases. Compliance will require organizations to move beyond merely identifying AI tools in use; they must also assess how these tools are used, what data is being shared, and what tasks the AI is performing. For instance, an employee using a generative AI tool to summarize sensitive internal documents introduces very different risks than someone using the same tool to draft marketing content.
For security and compliance leaders, the EU AI Act represents just one piece of a broader AI governance puzzle that will dominate 2025.
The next 12-18 months will require sustained focus and collaboration across security, compliance, and technology teams to stay ahead of these developments.
The Global Partnership on Artificial Intelligence (GPAI) is a multi-stakeholder initiative which aims to bridge the gap between theory and practice on AI by supporting cutting-edge research and applied activities on AI-related priorities.
https://gpai.ai/about/#:~:text=The%20Global%20Partnership%20on%20Artificial,activities%20on%20AI%2Drelated%20priorities.
290 Comments
Tomi Engdahl says:
Deepseek is way better in Python code generation than ChatGPT (talking about the “free” versions of both)
https://www.reddit.com/r/LocalLLaMA/comments/1i9txf3/deepseek_is_way_better_in_python_code_generation/
DeepSeek-Coder-V2: Breaking the Barrier of Closed-Source Models in Code Intelligence
https://github.com/deepseek-ai/DeepSeek-Coder-V2
Tomi Engdahl says:
Deepseek R1 Explained by a Retired Microsoft Engineer
https://www.youtube.com/watch?v=r3TpcHebtxM
Dave explains why Deepseek R1 is such a big deal, explains how it works, what’s new, and brings you up to date on the implications and fall out!
Tomi Engdahl says:
Run DeepSeek R1 Privately on Your Computer
https://www.youtube.com/watch?v=5kFV20LatL8
Tomi Engdahl says:
Master Local AI with DeepSeek-R1 In 10 Minutes
https://www.youtube.com/watch?v=CysTlwjx2vw
Learn how to use DeepSeek R1 locally to create your very own personal AI. This tutorial serves as the starting point for learning about DeepSeek R1 and the many things it’s capable of.
Tomi Engdahl says:
DeepSeek Blames Disruption on Cyberattack as Vulnerabilities Emerge
China’s DeepSeek blamed sign-up disruptions on a cyberattack as researchers started finding vulnerabilities in the R1 AI model.
https://www.securityweek.com/deepseek-blames-disruption-on-cyberattack-as-vulnerabilities-emerge/
Tomi Engdahl says:
Vlad Savov / Bloomberg:
Sam Altman says DeepSeek’s R1 is an “impressive model, particularly around what they’re able to deliver for the price” and OpenAI “will pull up some releases” — OpenAI Chief Executive Officer Sam Altman welcomed the debut of DeepSeek’s R1 model in a post on X late on Monday.
Sam Altman Praises DeepSeek R1 and Promises More From OpenAI
https://www.bloomberg.com/news/articles/2025-01-28/sam-altman-praises-deepseek-r1-and-promises-more-from-openai
OpenAI Chief Executive Officer Sam Altman welcomed the debut of DeepSeek’s R1 model in a post on X late on Monday.
The Chinese artificial intelligence startup that rocketed to global prominence has delivered an “impressive model, particularly around what they’re able to deliver for the price,” Altman wrote. Acknowledging DeepSeek as a competitor, Altman said it was “invigorating” and OpenAI will accelerate the release of some upcoming products.
Tomi Engdahl says:
Financial Times:
Bridgewater founder Ray Dalio says investor exuberance over AI has fuelled a “bubble” in US stocks that resembles the build-up to the dotcom bust in 1998-1999
Wall Street’s AI ‘bubble’ echoes dotcom excesses, Ray Dalio warns
https://www.ft.com/content/eef8dbc9-bd04-4502-bdc2-1092aa4251b2
Samantha Subin / CNBC:
Nvidia’s stock dropped 16.86%, closing at $118.58, losing nearly $600B in market cap, more than twice as much as any US company has ever lost in a single day — Nvidia lost close to $600 billion in market cap on Monday, the biggest drop for any company on a single day in U.S. history.
Nvidia sheds almost $600 billion in market cap, biggest one-day loss in U.S. history
https://www.cnbc.com/2025/01/27/nvidia-sheds-almost-600-billion-in-market-cap-biggest-drop-ever.html
Nvidia shares plunged 17% on Monday, resulting in a market cap loss of close to $600 billion, the biggest drop ever for a U.S. company.
The sell-off, which hit much of the U.S. tech sector, was sparked by concerns about increased competition from Chinese AI lab DeepSeek.
Data center companies that rely on Nvidia chips also plummeted, with Dell, Oracle and Super Micro Computer all falling by at least 8.7%.
Tomi Engdahl says:
Financial Times:
SemiAnalysis: DeepSeek has spent “well over $500M on GPUs over the history of the company”; TechInsights says it doesn’t see DeepSeek as “a big hit to Nvidia” — Short sellers profit as US chipmaker loses nearly $600bn in market value on Monday
https://www.ft.com/content/ee83c24c-9099-42a4-85c9-165e7af35105
Bloomberg:
DeepSeek says it used Nvidia H800 chips, available in China until October 2023, to train R1, suggesting future models could be hampered by US export controls
DeepSeek’s AI Model Tests Limits of US Restrictions on Nvidia Chips
https://www.bloomberg.com/news/articles/2025-01-27/deepseek-s-ai-model-tests-limits-of-us-curbs-on-nvidia-chips
Nvidia calls R1 an ‘excellent’ advance that meets US limits
Trump says release should be a ‘wake-up call’ for US companies
Tomi Engdahl says:
Analyysi: Kiinalaiset ajoivat ohi, eikä Trump keksinyt muuta kuin kehua heitä
https://yle.fi/a/74-20139740
Kiinalaisen Deepseek-tekoälyn käynnistämä hermoilu voi levitä Wall Streetiltä myös Valkoiseen taloon, kirjoittaa Yhdysvaltain-kirjeenvaihtaja Ilmari Reunamäki.
WASHINGTON Tekoälyn aikakauden Sputnik-hetki, toisteltiin eilen amerikkalaisilla tv-kanavilla kiinalaisen Deepseekin aiheuttamaa yllättävää huolta.
Sputnik oli Neuvostoliiton vuonna 1957 avaruuteen laukaisema maailman ensimmäinen satelliitti. Yhdysvallat oli siihen asti uskonut olevansa edellä avaruuskilvassa, mutta ei ollutkaan.
Eilen Deepseek kaatoi Yhdysvalloissa vallinneen uskomuksen, että Kiina oli jäänyt ratkaisevasti jälkeen tekoälykilvassa.
Deepseek ei ole Sputnikin tapaan mullistava siinä, mitä se osaa tehdä. Se on käyttäjäkokemukseltaan hyvin samankaltainen kuin tunnettu Chat GPT -tekoäly.
Taika piilee kuitenkin siinä, kuinka kevyesti Deepseek pyörii ja kuinka yritys mainostaa pystyvänsä tarjoamaan Chat GPT:ta vastaavan palvelun yli kymmenen kertaa halvemmalla.
Suhteessa pieni kiinalainen yritys kertoi myös kouluttaneensa tekoälynsä halvalla ilman Nvidian uusimpia grafiikkapiirejä. Kiinalaiset osoittivat amerikkalaisille, että pakon edessä luovuudella voi säästää pitkän pennin.
Tomi Engdahl says:
Deepseek-tekoälyn takaa löytyy 40-vuotias sijoittajanörtti, jonka sivuharrastus heiluttaa nyt maailman pörssejä
https://yle.fi/a/74-20139825
Maailmaa valloittavan Deepseek-yhtiön perustaja Liang Wenfeng ei vastaa käsitystä tekoälypioneerista. Aluksi hänellä olikin uskottavuusongelma.
Kiinalainen tekoäly-yhtiö Deepseek järisytti ennennäkemättömällä tavalla maanantaina pörssejä ja tekoälyalaa ympäri maailmaa. Syynä oli yhtiön kuin vaivihkaa asiantuntijoiltakin julkaisema uusi tekoälymalli.
Myös Deepseek-tekoälyn perustajasta Liang Wenfengista maailma tietää vasta vähän. Kuvat ja videot 40-vuotiaasta sijoittajasta ovat tiukassa.
Haastatteluja Liang on itse antanut vähän. Liikekumppanit ja kilpailijat kommentoivat häntä yleensä nimettömästi.
Oman Wikipedia-sivunsakin hän sai vasta viime viikolla.
Viime viikon maanantaina Liang putkahti esiin Kiinan valtiontelevisiossa, kun hän ainoana tekoälyalan edustajana osallistui Kiinan pääministerin Li Qiangin isännöimään liike-elämän tapaamiseen.
Kuvissa pääministeriä vastapäätä istuu harmaaseen pukuun sonnustautunut, ikäistään nuoremmalta näyttävä mies, jota Li näyttää kuuntelevan tarkasti.
– Keskittykää teknologian avainalojen läpimurtoon, pääministeri kehotti osallistujia.
Tämän viikon maanantaina Liang järkyttikin omalla läpimurrollaan niin Piilaakson tekoälypiirejä kuin Wall Streetin sijoittajiakin.
Kiina puolestaan oli saanut uuden kansallisen sankarin.
Liangin visio luovuudesta johtolankana
Toissa vuonna Liang perusti Deepseekin. Tavoite oli kehittää tekoäly, joka vastaa ihmisen älykkyyttä.
Deepseek on pieni yhtiö, jonka pääkonttori sijaitsee Kiinan itäosassa, Hangzhoussa. Yhtiö työllistää tiettävästi parisataa ihmistä Hangzhoun ja Pekingin toimistoissaan.
Liang on pystynyt rahoittamaan yhtiötä itse, eikä sen ole toistaiseksi tarvinnut kerätä markkinoilta ulkopuolista rahoitusta.
Liangin sanotaan nyt omistautuneen sille, että hän kasvattaa Deepseekista tekoälyalan Kiinasta ponnistavan johtajan.
Viime vuonna The China Academy -julkaisulle antamassa haastattelussaan Liang totesi, että Kiinan tapa peesata ulkomailla tehtyjä keksintöjä ei ole pitkän päälle kestävää.
Siksi hän haluaa yhtiönsä olevan enemmän kuin rahantekokone.
– Tavoitteemme ei ole tehdä nopeasti tulosta vaan edistää eturintamassa koko alan kasvua, Ling sanoi.
Kiinasta kuuluu yllätysuutisia, jotka järisyttävät tekoälykehitystä ja kilpailua maailmanherruudesta
https://yle.fi/a/74-20139427
Yhdysvaltain teknologiasanktioiden teho hiipuu. Osin rajoitukset ovat kääntyneet itseään vastaan.
Lähes tuntematon kiinalainen tekoäly-yhtiö Deepseek on aiheuttanut valtavan kuhinan Piilaaksossa.
Yhtiö julkisti alkuviikolla uuden tekoälymalli R1:n monimutkaiseen ongelmanratkaisuun. Se on teknologisesti hyvin lähellä yhdysvaltalaisia tekoälymalleja ja päihittää ne tietyissä toiminnoissa.
Esimerkiksi pääomasijoittaja Marc Andreessen, joka on toiminut Yhdysvaltain presidentin Donald Trumpin teknologianeuvonantajana, sanoi kiinalaistekoälyn kehitystä ällistyttäväksi Wall Street Journalissa.
– Deepseek R1 on yksi hämmästyttävimpiä ja vaikuttavimpia teknologisia läpimurtoja, mitä olen koskaan nähnyt.
Uutiset Kiinasta huolettavat Yhdysvalloissa ainakin kolmesta syistä.
Yhdysvaltain teknologinen etumatka hiipuu
Tekoälykehityksen kärkeä ovat pitäneet selvällä erolla yhdysvaltalaiset yhtiöt – osin Yhdysvaltain hallinnon avulla.
Deepseekin läpimurrolla Kiina kuroo Yhdysvaltain etumatkaa kiinni jättiharppauksella. Jo joulukuussa sama yhtiö julkisti huipputehokkaan kielimallin eli ChatGPT:n kaltaisen tekoälytyökalun. The Economistin mukaan sen teholle vetävät vertoja vain Googlen ja ChatGPT:n kehittäneen OpenAI:n kielimallit.
Tekoälysanktiot tehoavat heikosti
Yhdysvallat ja Kiina ovat olleet kauppasodassa vuodesta 2018 lähtien, jolloin Trump asetti ensimmäiset järeät tullit.
Tekoälyn saralla Yhdysvallat on yrittänyt jarruttaa Kiinan kehitystä useilla sanktioilla ja kieltämällä kehittyneiden tekoälysirujen viennin maahan. Esimerkiksi maailman arvokkain yhtiö, siruvalmistaja Nvidia joutuu tuottamaan Kiinan markkinoille heikennettyä versiota tekoälysiruistaan.
Se, että Deepseek onnistui luomaan huipputehokkaan tekoälyn rajoituksista huolimatta, osoittaa Yhdysvaltain sanktioiden tehon hiipuvan.
Mitä tekoälykehityksessä on pelissä?
Facebookin ja Instagramin emoyhtiö Meta kertoo käyttävänsä tekoälyinvestointeihin yksin tänä vuonna Reutersin mukaan yli 60 miljardia.
Esimerkiksi se ja Trumpin julkistama 500 miljardin hanke kertovat siitä, kuinka mittavia panokset tekoälykehityksessä ovat. Tekoälystä odotetaan valtavasti lisää tuottavuus- ja talouskasvua, kun sitä aletaan käyttää laajamittaisesti.
Vanhan viisauden mukaan teknologian lyhyen aikavälin vaikutukset yliarvioidaan ja pitkän aikavälin vaikutukset aliarvioidaan.
Näin on luultavasti myös tekoälyn kanssa. Se tuskin mullistaa taloutta vielä vuosiin, mutta pidemmän ajan saatossa sen vaikutukset voivat myllertää suurta osaa toimialoista, tuotannosta ja ammateista.
Eikä kyse ole pelkästä taloudesta. Tekoälystä odotetaan jättäharppauksia myös sotateknologiaan.
Näistä syistä maailmanmahdit Kiina ja Yhdysvallat kilpailevat niin kovaa tekoälykehityksessä. Sillä, kenellä on käytössään paras teknologia, on suuri etumatka taloudessa, puolustuksessa ja globaalissa vaikutusvallassa.
Tomi Engdahl says:
Voiko ChatGPT parantaa elämääsi? Toimittaja testasi vaikuttajien ylistämät tekoälypromptit
Tekoäly on noussut arjen keskeiseksi puheenaiheeksi. Toimittaja Roosa Laksio testasi, mitkä promptit oikeasti auttavat ja mitkä ovat pelkkää hypetystä.
https://yle.fi/a/74-20124118
Tomi Engdahl says:
Microsoft just renamed Office to Microsoft 365 Copilot on Windows 11 for everyone
https://www.windowslatest.com/2025/01/18/microsoft-just-renamed-office-to-microsoft-365-copilot-on-windows-11-for-everyone/
Once again, Microsoft has proven its knack for rebranding and sometimes complicating its product lineup. They recently announced that they’ll be rebranding the Microsoft 365 app to Microsoft 365 Copilot. And today, we noticed that “Microsoft 365 Office” is now called “Microsoft 365 Copilot” after the update.
Microsoft’s rebranding efforts are now beyond control. We went from Office > Office 365 > Microsoft 365 Office to Microsoft 365 Copilot
Microsoft previously said that this change aligns with the company’s goal to make AI an integral part of its ecosystem. And with Copilot in the focus, Microsoft is moving forward with its AI-first strategy.
The Microsoft 365 Copilot rebrand is now rolling out on Windows 11 devices. According to Microsoft, doing so was necessary as the current naming structure is confusing. The only solution, according to them, was to add the “Copilot” tag to every Enterprise product.
They are also changing the Microsoft 365 Copilot URL to M365Copilot.com and redirecting office.com and Microsoft365.com to m365.cloud.microsoft.
Windows Latest recently spotted a roadmap and saw an idea of what’s coming with the rebrand. Some major changes include:
Upon clicking the Copilot keyboard key in the Microsoft 365 app, the Copilot page will automatically launch.
The top header has been retired, giving the UI a cleaner look. This will migrate all the account-related options to the bottom of the left side menu.
The organization name header is also being retired; this used to appear in the top header.
All the AI features, including Microsoft 365 Copilot Chat and Copilot Pages, will be moved to the left sidebar, and a Back button will be added.
The search bar will appear on the homepage, allowing you to search for content across different sections.
All these changes are applicable to both the web and desktop versions of the newly rebranded Microsoft 365 Copilot. As of now, everything is live, and these changes shall be reflected or will soon be on your device.
As soon as you open the new Microsoft 365 Copilot app, it will greet you with a “Welcome to Microsoft 365 Copilot” pop-up, and then you’ll land on the homepage.
Below Home, we’ve the Copilot button, which simply loads copilot.microsoft.com but is more deeply integrated into the Microsoft 365 ecosystem, so you can ask it
Was this rebrand required?
Probably not, but Microsoft has always had trouble sticking to a name for long. Previously, Copilot was Bing Chat, and now Microsoft has changed the Microsoft 365 app to Microsoft 365 Copilot.
For users with an Entra account, Microsoft Copilot has been renamed to Microsoft 365 Copilot Chat. This name is too long and was definitely not needed.
Look at the above screenshot of the taskbar with two apps open – Microsoft 365 Copilot and Copilot. Which one is what? Of course, if you’re a tech-savvy person, you might understand it easily, but someone who is new to Microsoft and Windows ecosystems can be confused.
Worse, Microsoft 365 Copilot’s Copilot section can do pretty much everything the dedicated Copilot app can.
The rebrand of the Microsoft 365 app to Microsoft 365 Copilot was unnecessary, but here we are.
Tomi Engdahl says:
I Tried AI to Summarize a Book. The Plagiarism Protections Are Working https://cnet.co/3EdRYjY
Tomi Engdahl says:
DeepSeek, the viral AI company, has released a new set of multimodal AI models that it claims can outperform OpenAI’s DALL-E 3.
The models, which are available for download from the AI dev platform Hugging Face, are part of a new model family that DeepSeek is calling Janus-Pro. They range in size from 1 billion to 7 billion parameters.
Parameters roughly correspond to a model’s problem-solving skills, and models with more parameters generally perform better than those with fewer parameters.
Read more from Kyle Wiggers on Janus-Pro here: https://tcrn.ch/40Bc5Qm
#TechCrunch #technews #artificialintelligence #DeepSeek #OpenAI
Tomi Engdahl says:
Scientists are very concerned about artificial intelligence (AI) because frontier AI systems have surpassed what they are calling “the self-replicating red line.” The scientists say that such an advancement “is an early signal for rogue AIs.” .. . is it time to get serious or is this much ado about nothing?
Scientists Are Sounding An Alarm Because AI Has Learned How To Self-Replicate
https://brobible.com/culture/article/ai-learned-how-to-self-replicate/
Scientists are very concerned about artificial intelligence (AI) because frontier AI systems have surpassed what they are calling “the self-replicating red line.” The scientists say that such an advancement “is an early signal for rogue AIs.”
“Successful self-replication under no human assistance is the essential step for AI to outsmart the human beings,”
As part of their research, the scientists discovered two AI systems driven by Meta’s Llama31-70B-Instruct and Alibaba’s Qwen25-72B-Instruct “have already surpassed the self-replicating red line. In 50% and 90% experimental trials, they succeed in creating a live and separate copy of itself respectively.”
Why is this a concern? Because, as they state in their research paper, that means AI systems could use “self-replication to avoid shutdown and create a chain of replica to enhance the survivability, which may finally lead to an uncontrolled population of AIs.”
Why is this a concern? Because, as they state in their research paper, that means AI systems could use “self-replication to avoid shutdown and create a chain of replica to enhance the survivability, which may finally lead to an uncontrolled population of AIs.”
Tomi Engdahl says:
Chinese-made DeepSeek AI model records extensive online user data, stores it in China-based servers
News
By Matthew Connatser published 13 hours ago
What doesn’t DeepSeek collect?
https://www.tomshardware.com/tech-industry/artificial-intelligence/chinese-made-deepseek-ai-model-collects-extensive-user-data-stores-it-on-china-based-servers?fbclid=IwZXh0bgNhZW0CMTEAAR3JCWWKZC8KYvXwk9B8nRgVmykc6SAgrzJvEraRRuDpfuznKVPToyVjsic_aem_4rNAka-0t_7SXsRdfbhCDQ
DeepSeek’s newest R1 large language model has already become notorious after its release cratered AI stocks, and revelations about its privacy policy might raise eyebrows even more — the company records extensive data from its online users, including keystrokes, passwords, and data entered in queries like images and text, and then stores it in China-based servers.
Personal information, including date of birth, email addresses, phone numbers, and passwords, are all fair game, according to DeepSeek. Any content users give to the R1 LLM, from text and audio prompts to uploaded files, may also be collected by DeepSeek. And whenever someone contacts DeepSeek, it says it might keep users’ proof of identity, which presumably means documents like a driver’s license.
But that’s not all. DeepSeek records anything related to users’ hardware: IP addresses, phone models, language, etc. Its collection efforts are so thorough that the company notes “keystroke patterns or rhythms.” Cookies, a classic method of tracking users on the Internet, also contribute to user data collection.
Because R1 is ‘open source,’ it can be run anywhere on any hardware, which is generally good for privacy — running the model locally on your own hardware will presumably not lead to data collection. However, DeepSeek offers online access to R1 via its website and mobile app, which means the AI company handles and stores online users’ data. Thankfully, DeepSeek is very transparent about what data it collects from online users, where it’s stored, and what it does with it. It details it all in its privacy policy webpage, which reveals that there’s almost nothing the company doesn’t collect.
As for where all this information is stored, the privacy policy says it’s all kept inside servers located in China, a point that has the potential to spark serious controversy. Concerns about the personal details of Americans being in the hands of the Chinese government was a key factor in the Biden administration’s attempt to ban TikTok, raising the possibility that DeepSeek might come under similar scrutiny.
https://chat.deepseek.com/downloads/DeepSeek%20Privacy%20Policy.html
Tomi Engdahl says:
Markkinat menivät sekaisin – Mitä tekee Trump?
Akatemiatutkija Matti Ylönen uskoo, että Donald Trumpilta odotetaan teknologiateollisuuden tukemista, mutta se on toinen asia, mitä oikeasti on edes tehtävissä.
https://www.iltalehti.fi/ulkomaat/a/de19ccf7-58be-4549-9f79-6a6ac8570b37
Helsingin yliopiston akatemiatutkija Matti Ylönen uskoo, että yhdysvaltalaiset teknologiajätit odottavat presidentti Donald Trumpilta etujensa ajamista.
– Kyllä varmasti sellainen odotus on, kun oikeastaan kaikki suurimmat teknologiafirmat satsasivat isosti Trumpin kampanjaan.
Monen yhtiön edut ovat juuri nyt vaarassa, kun erityisesti tekoälybisneksen ympärillä kuohuu.
Kiinalainen tekoälysovellus Deepseek ilmestyi kuin tyhjästä ja singahti latauslistojen kärkeen. Tapahtunut ravistelee Yhdysvaltojen teknologiamarkkinoita hurjalla tavalla.
Pahimmin siipeensä on saanut Nvidia, jonka suorittimia käytetään tekoälytoiminnoissa. Deepseek vaikuttaa osoittaneen, että tekoälyn kehittäminen ei vaadikaan älytöntä suoritinsatsausta, ja Nvidian arvosta suli liki hetkessä lähes 600 miljardia dollaria.
Myös muiden tekoälybisnekseen kytkeytyvien yhtiöiden kurssit notkahtivat.
”Ei millään kikalla”
Trumpilta ehkä odotetaan toimia, mutta kokonaan toinen juttu on se, mitä ylipäätään on tehtävissä.
– Jos joku nyt kehittää aikaisempaa huomattavasti tehokkaamman ja paremman systeemin kielimallien tai tekoälyn treenaamiseen, niin eihän siihen millään kikalla pystytä vastaamaan, Matti Ylönen sanoo.
Pitkäjänteisyyttä
Matti Ylönen toteaa, että vaikka markkinat reagoivatkin nopeasti, Deepseekin vaikutukset näkyvät vasta pidemmällä aikavälillä.
Notkahduksesta huolimatta yhdysvaltalaisten teknologiajättien bisnes ei ole ollenkaan välttämättä lähtemässä alta.
– Teknologisen kapasiteetin rakentaminen on tietysti hyvin pitkäjänteistä, ja Yhdysvalloissahan sitä on pitkäjänteisesti tehtykin valtion rahalla, Matti Ylönen sanoo.
Tekoälybisneksessä liikkuvat valtavat rahat, ja Donald Trump ymmärtää sen.
Hän ilmoitti viime viikolla, liki välittömästi toisen kautensa aloitettuaan, jättimäisestä Stargate- eli Tähtiportti-hankkeesta. Trump julisti, että jopa 500 miljardin dollarin tekoälyhankkeen odotetaan luovan Yhdysvaltoihin 100 000 työpaikkaa.
– Hänellähän on isot puheet, mutta saa nähdä, mitä tämä tulee käytännössä tarkoittamaan, Matti Ylönen sanoo.
Hankkeen rahoitusta ja vaikutuksia on epäilty, mutta hanke osoittaa joka tapauksessa, että Yhdysvallat ja Trump kokevat tekoälyliiketoiminnan tärkeäksi asiaksi.
Kun Deepseek sinkoutui vain muutamia päiviä Tähtiportti-julistuksen jälkeen horjuttamaan markkinoita, Trump innostui kehumaan aikaisemmin tuntemattoman kiinalaisfirman kekseliäisyyttä.
Trumpin mukaan on myönteistä, että on onnistuttu luomaan menetelmä, joka mahdollistaa tekoälyn kehittämisen huomattavasti aiempaa halvemmalla.
Trump sanoi toivovansa, että Deepseekin tulo toimii yhdysvaltalaisille toimijoille herätyksenä ja että ne panostavat kilpailuun. Trump kertoi kuitenkin uskovansa, että Yhdysvaltojen asema alan kärkimaana ei ole vaarassa.
Tomi Engdahl says:
Financial Times:
OpenAI says it had seen some evidence that DeepSeek used “distillation” to train its open-source competitor by using outputs from OpenAI’s proprietary models — White House AI tsar David Sacks raises possibility of alleged intellectual property theft
OpenAI says it has evidence China’s DeepSeek used its model to train competitor
White House AI tsar David Sacks raises possibility of alleged intellectual property theft
https://www.ft.com/content/a0dfedd1-5255-4fa9-8ccc-1fe01de87ea6?accessToken=zwAGLNJX-fBAkdOg3-3RUlVPqdOMzB_gHeh-pg.MEYCIQCGjo04z0mtOsKbDspQLq2BMXyw8SbQnlYePOuqiqr6QgIhAInK67eBkYuZS-77ljnP-y–EJdN1wwRQ8GIR8sKMFgE&sharetype=gift&token=1eebbaa7-a4e6-4251-b665-c2f2562b38e4
OpenAI says it has found evidence that Chinese artificial intelligence start-up DeepSeek used the US company’s proprietary models to train its own open-source competitor, as concerns grow over a potential breach of intellectual property.
The San Francisco-based ChatGPT maker told the Financial Times it had seen some evidence of “distillation”, which it suspects to be from DeepSeek.
The technique is used by developers to obtain better performance on smaller models by using outputs from larger, more capable ones, allowing them to achieve similar results on specific tasks at a much lower cost.
Distillation is a common practice in the industry but the concern was that DeepSeek may be doing it to build its own rival model, which is a breach of OpenAI’s terms of service.
“The issue is when you [take it out of the platform and] are doing it to create your own model for your own purposes,” said one person close to OpenAI.
OpenAI declined to comment further or provide details of its evidence. Its terms of service state users cannot “copy” any of its services or “use output to develop models that compete with OpenAI”.
DeepSeek’s release of its R1 reasoning model has surprised markets, as well as investors and technology companies in Silicon Valley. Its built-on-a-shoestring models have attained high rankings and comparable results to leading US models.
Tomi Engdahl says:
promptfoo:
DeepSeek-R1 refuses to answer ~85% of 1,360 prompts on sensitive topics in China, but the restrictions can be bypassed via simple jailbreaking — DeepSeek-R1 is a blockbuster open-source model that is now at the top of the U.S. App Store. — As a Chinese company, DeepSeek is beholden to CCP policy.
https://www.promptfoo.dev/blog/deepseek-censorship/
Tomi Engdahl says:
Jackie Davalos / Bloomberg:
David Sacks says there’s “substantial evidence” that DeepSeek “distilled knowledge out of OpenAI models and I don’t think OpenAI is very happy about this” — White House artificial intelligence czar David Sacks said there’s “substantial evidence” …
DeepSeek Leaned on OpenAI Models, White House AI Czar Sacks Says
https://www.bloomberg.com/news/articles/2025-01-28/ai-czar-sacks-says-evidence-deepseek-leaned-on-openai-s-models
Tomi Engdahl says:
Hayden Field / CNBC:
The US Navy has instructed its members to avoid using DeepSeek “in any capacity” due to “potential security and ethical concerns” — The U.S. Navy has instructed its members to avoid using artificial intelligence technology from China’s DeepSeek, CNBC has learned.
U.S. Navy bans use of DeepSeek due to ‘security and ethical concerns’
https://www.cnbc.com/2025/01/28/us-navy-restricts-use-of-deepseek-ai-imperative-to-avoid-using.html
The U.S. Navy issued a warning to its members to avoid using DeepSeek “in any capacity,” due to “potential security and ethical concerns.”
The warning was sent out on Friday as buzz about the Chinese artificial intelligence startup was picking up across the tech industry.
The email instructed all team members not to use DeepSeek “for any work-related tasks or personal use.”
Tomi Engdahl says:
Bloomberg:
Sources: Microsoft and OpenAI are investigating whether data output from OpenAI’s API was obtained in an unauthorized manner by a group linked to DeepSeek — Microsoft’s security researchers in the fall observed individuals they believe may be linked to DeepSeek exfiltrating a large amount …
Microsoft Probing If DeepSeek-Linked Group Improperly Obtained OpenAI Data
https://www.bloomberg.com/news/articles/2025-01-29/microsoft-probing-if-deepseek-linked-group-improperly-obtained-openai-data
Tomi Engdahl says:
Radhika Rajkumar / ZDNET:
Block introduces an on-device, open-source AI agent called Goose, which allows developers to choose their preferred LLM to automate engineering tasks — Whether for developer needs or mundane tasks, the artificial intelligence (AI) tide appears to be turning in favor of open-source solutions.
Block’s new open-source AI agent ‘goose’ lets you change direction mid-air
Block built its agent – ‘codename goose’ – to do it all, from writing code to ordering your dinner. Here’s how to access it.
https://www.zdnet.com/article/blocks-new-open-source-ai-agent-goose-lets-you-change-direction-mid-air/
Tomi Engdahl says:
Kyle Wiggers / TechCrunch:
Hugging Face researchers unveil Open-R1, a project to “systematically reconstruct DeepSeek-R1′s data and training pipeline” for the open-source community — Barely a week after DeepSeek released its R1 “reasoning” AI model — which sent markets into a tizzy …
Hugging Face researchers are trying to build a more open version of DeepSeek’s AI ‘reasoning’ model
https://techcrunch.com/2025/01/28/hugging-face-researchers-are-trying-to-build-a-more-open-version-of-deepseeks-ai-reasoning-model/
Barely a week after DeepSeek released its R1 “reasoning” AI model — which sent markets into a tizzy — researchers at Hugging Face are trying to replicate the model from scratch in what they’re calling a pursuit of “open knowledge.”
Hugging Face head of research Leandro von Werra and several company engineers have launched Open-R1, a project that seeks to build a duplicate of R1 and open source all of its components, including the data used to train it.
The engineers said they were compelled to act by DeepSeek’s “black box” release philosophy. Technically, R1 is “open” in that the model is permissively licensed, which means it can be deployed largely without restrictions. However, R1 isn’t “open source” by the widely accepted definition because some of the tools used to build it are shrouded in mystery. Like many high-flying AI companies, DeepSeek is loathe to reveal its secret sauce.
“The R1 model is impressive, but there’s no open dataset, experiment details, or intermediate models available, which makes replication and further research difficult,” Elie Bakouch, one of the Hugging Face engineers on the Open-R1 project, told TechCrunch. “Fully open sourcing R1’s complete architecture isn’t just about transparency — it’s about unlocking its potential.”
R1 broke into the mainstream consciousness after DeepSeek’s chatbot app, which provides free access to R1, rose to the top of the Apple App Store charts. The speed and efficiency with which R1 was developed — DeepSeek released the model just weeks after OpenAI released o1 — has led many Wall Street analysts and technologists to question whether the U.S. can maintain its lead in the AI race.
The Open-R1 project is less concerned about U.S. AI dominance than “fully opening the black box of model training,” Bakouch told TechCrunch. He noted that, because R1 wasn’t released with training code or training instructions, it’s challenging to study the model in depth — much less steer its behavior.
“Having control over the dataset and process is critical for deploying a model responsibly in sensitive areas,” Bakouch said. “It also helps with understanding and addressing biases in the model. Researchers require more than fragments … to push the boundaries of what’s possible.”
Steps to replication
The goal of the Open-R1 project is to replicate R1 in a few weeks, relying in part on Hugging Face’s Science Cluster, a dedicated research server with 768 Nvidia H100 GPUs.
The Hugging Face engineers plan to tap the Science Cluster to generate datasets similar to those DeepSeek used to create R1. To build a training pipeline, the team is soliciting help from the AI and broader tech communities on Hugging Face and GitHub, where the Open-R1 project is being hosted.
“We need to make sure that we implement the algorithms and recipes [correctly,]” von Werra told TechCrunch, “but it’s something a community effort is perfect at tackling, where you get as many eyes on the problem as possible.”
There’s a lot of interest already. The Open-R1 project racked up 10,000 stars in just three days on GitHub. Stars are a way for GitHub users to indicate that they like a project or find it useful.
If the Open-R1 project is successful, AI researchers will be able to build on top of the training pipeline and work on developing the next generation of open source reasoning models, Bakouch said. He hopes the Open-R1 project will yield not only a strong open source replication of R1, but also a foundation for better models to come.
“Rather than being a zero-sum game, open source development immediately benefits everyone, including the frontier labs and the model providers, as they can all use the same innovations,” Bakouch said.
Open-R1: a fully open reproduction of DeepSeek-R1
https://huggingface.co/blog/open-r1
What is DeepSeek-R1?
If you’ve ever struggled with a tough math problem, you know how useful it is to think a little longer and work through it carefully. OpenAI’s o1 model showed that when LLMs are trained to do the same—by using more compute during inference—they get significantly better at solving reasoning tasks like mathematics, coding, and logic.
However, the recipe behind OpenAI’s reasoning models has been a well kept secret. That is, until last week, when DeepSeek released their DeepSeek-R1 model and promptly broke the internet (and the stock market!).
Besides performing as well or better than o1, the DeepSeek-R1 release was accompanied by a detailed tech report that outlined the key steps of their training recipe. This recipe involved several innovations, most notably the application of pure reinforcement learning to teach a base language model how to reason without any human supervision. As shown in the figure below, making a powerful reasoning model is now very simple if you have access to a capable base model and a high-quality data mixture:
However, the DeepSeek-R1 release leaves open several questions about:
Data collection: How were the reasoning-specific datasets curated?
Model training: No training code was released by DeepSeek, so it is unknown which hyperparameters work best and how they differ across different model families and scales.
Scaling laws: What are the compute and data trade-offs in training reasoning models?
These questions prompted us to launch the Open-R1 project, an initiative to systematically reconstruct DeepSeek-R1’s data and training pipeline, validate its claims, and push the boundaries of open reasoning models. By building Open-R1, we aim to provide transparency on how reinforcement learning can enhance reasoning, share reproducible insights with the open-source community, and create a foundation for future models to leverage these techniques.
How did they do it?
DeepSeek-R1 is a reasoning model built on the foundation of DeepSeek-V3. Like any good reasoning model, it starts with a strong base model, and DeepSeek-V3 is exactly that. This 671B Mixture of Experts (MoE) model performs on par with heavyweights like Sonnet 3.5 and GPT-4o. What’s especially impressive is how cost-efficient it was to train—just $5.5M—thanks to architectural changes like Multi Token Prediction (MTP), Multi-Head Latent Attention (MLA) and a LOT (seriously, a lot) of hardware optimization.
DeepSeek also introduced two models: DeepSeek-R1-Zero and DeepSeek-R1, each with a distinct training approach. DeepSeek-R1-Zero skipped supervised fine-tuning altogether and relied entirely on reinforcement learning (RL), using Group Relative Policy Optimization (GRPO) to make the process more efficient. A simple reward system was used to guide the model, providing feedback based on the accuracy and structure of its answers. This approach helped the model develop useful reasoning skills, such as breaking problems into steps and verifying its own outputs. However, its responses often lacked clarity and were difficult to read.
That’s where DeepSeek-R1 comes in. It started with a “cold start” phase, fine-tuning on a small set of carefully crafted examples to improve clarity and readability. From there, it went through more RL and refinement steps, including rejecting low-quality outputs with both human preference based and verifiable reward, to create a model that not only reasons well but also produces polished and consistent answers.
Open-R1: the missing pieces
The release of DeepSeek-R1 is an amazing boon for the community, but they didn’t release everything—although the model weights are open, the datasets and code used to train the model are not
The goal of Open-R1 is to build these last missing pieces so that the whole research and industry community can build similar or better models using these recipes and datasets. And by doing this in the open, everybody in the community can contribute!
As shown in the figure below, here’s our plan of attack:
Step 1: Replicate the R1-Distill models by distilling a high-quality reasoning dataset from DeepSeek-R1.
Step 2: Replicate the pure RL pipeline that DeepSeek used to create R1-Zero. This will involve curating new, large-scale datasets for math, reasoning, and code.
Step 3: Show we can go from base model → SFT → RL via multi-stage training.
Tomi Engdahl says:
Block’s new open-source AI agent ‘goose’ lets you change direction mid-air
Block built its agent – ‘codename goose’ – to do it all, from writing code to ordering your dinner. Here’s how to access it.
https://www.zdnet.com/article/blocks-new-open-source-ai-agent-goose-lets-you-change-direction-mid-air/
Tomi Engdahl says:
Former OpenAI safety researcher brands pace of AI development ‘terrifying’
Steven Adler expresses concern industry taking ‘very risky gamble’ and raises doubts about future of humanity
https://www.theguardian.com/technology/2025/jan/28/former-openai-safety-researcher-brands-pace-of-ai-development-terrifying
Tomi Engdahl says:
Kyle Wiggers / TechCrunch:
Hugging Face launches Inference Providers, which makes it easier for developers to run AI models on 3rd-party clouds; launch partners include SambaNova and Fal
Hugging Face makes it easier for devs to run AI models on third-party clouds
https://techcrunch.com/2025/01/28/hugging-face-makes-it-easier-for-devs-to-run-ai-models-on-third-party-clouds/
AI dev platform Hugging Face has partnered with third-party cloud vendors, including SambaNova, to launch Inference Providers, a feature designed to make it easier for devs on Hugging Face to run AI models using the infrastructure of their choice.
Other partners involved with the new effort include Fal, Replicate, and Together AI.
Hugging Face says its partners have worked with it to build access to their respective data centers for running models into Hugging Face’s platform. Now, developers on Hugging Face can, for example, spin up a DeepSeek model on SambaNova’s servers from a Hugging Face project page in just a few clicks.
Hugging Face has long offered its own in-house solution for running AI models. But in a blog post Tuesday, the company explained that its focus has shifted to collaboration, storage, and model distribution capabilities.
https://huggingface.co/blog/inference-providers
Tomi Engdahl says:
Kiinalainen tekoälyfirma iski kuin tyhjästä – Yhdysvalloissa voimakas reaktio
Kiinalaisesta tekoälymallista on kohistu paljon. Siihen on kaksi syytä.
https://www.iltalehti.fi/digiuutiset/a/483a4d23-ccd5-4671-8587-906d82e853cb
Kiinalaisen tekoälystartupin Deepseekin sovellus nousi Applen sovelluskaupan ladatuimmaksi sovellukseksi Yhdysvalloista.
Asiasta kertoo uutistoimisto Reuters.
Yhtiö julkaisi sovelluksensa viime viikon maanantaina ja nousi teknologiamaailman yhdeksi suurimmaksi puheenaiheeksi. Sovellus on maailman edistyksellisimpänä kielimallina pidetyn ChatGPT:n kaltainen tekoälytuote.
R1-nimellä tunnettu kielimalli sai hehkutusta jopa Yhdysvaltojen hallinnon läheltä. Pääomasijoittaja ja presidentti Donald Trumpin neuvonantaja Marc Andreessen sanoi X:ssä mallin olevan yksi vakuuttavimmista läpimurroista joita hän on koskaan nähnyt.
– Avoimen lähdekoodin sovelluksena se on lahja maailmalle, Andreessen kirjoitti.
The Wall Street Journalin mukaan Deepseekin tekoälymallit olivat Berkeleyn yliopiston pitämässä listauksessa tehokkaampia kuin esimerkiksi Elon Muskin omistaman X:n käyttämä Grok. Listauksessa Googlen Gemini-malli on ykkönen.
Markkinoilla teknologiaindeksin Nasdaq 100 -futuurit laskivat 2,6 prosenttia ja S&P 500 -indeksin futuurit laskivat 1,5 prosenttia Deepseekin menestyksen jälkeen, kertoo uutistoimisto Reuters.
Kaksi huomiota
Deepseek herättää huomiota kahdesta syystä.
Ensinnäkin R1-malli on kehitetty todella halvalla. The Wall Street Journalin mukaan yhtiö kertoo, että mallin rakentaminen maksoi vaivaiset 5,6 miljoonaa dollaria, eli noin 5,3 miljoonaa euroa.
Hintaa voi verrata vaikkapa tekoälykehittäjä Anthropiciin, jonka kehittämän mallin hintalappu liikkuu 100 miljoonasta dollarista aina miljardiin dollariin asti.
Kustannustehokkuus johtuu siitä, että Reutersin mukaan yhtiö kertoo kouluttaneensa mallin Nvidian H800-siruilla, jotka eivät ole suorituskyvyiltään alan huippua.
Toiseksi Deepseekissä huomionarvoista on, että se perustuu avoimeen lähdekoodiin. Somejätti Metan tekoälypäällikkö Yann LeCun sanoi Fox Businessin mukaan, että avoin lähdekoodi on yhtiön mallissa paljon tärkeämpää kuin se, voittaako kiinalaistekoäly amerikkalaisen kilpailijansa.
– Deepseek on hyödyntänyt avointa tutkimusta ja avoimen lähdekoodin tuotteita ja rakentaneet ideansa muiden työn päälle. Koska heidän työnsä on julkista, kaikki voivat hyötyä siitä, LeCun kirjoitti Fox Businessin mukaan.
DeepSeek hit by cyberattack as users flock to Chinese AI startup
https://www.reuters.com/technology/artificial-intelligence/chinese-ai-startup-deepseek-overtakes-chatgpt-apple-app-store-2025-01-27/
Tomi Engdahl says:
Mikko Hyppöseltä pelottava visio uudesta keksinnöstä: ”Sitten se kyllä kertoo sinulle…”
Maailmaa kohauttanut keksintö uhkaa antaa avaimet mitä vaarallisimpaan tietoon.
Mikko Hyppöseltä pelottava visio uudesta keksinnöstä: ”Sitten se kyllä kertoo sinulle…”
https://www.is.fi/digitoday/tietoturva/art-2000010994491.html
Lue tiivistelmä
Kiinalainen Deepseek-tekoäly herättää huolta avoimen lähdekoodinsa vuoksi.
Deepseek voi mahdollistaa vaarallisen tiedon, kuten vetypommin rakentamisen, levittämisen.
Tietoturva-asiantuntija Mikko Hyppönen uskoo, että Deepseek voi auttaa rikollisia löytämään uusia haavoittuvuuksia.
Deepseekin avoimuus uhkaa tehdä tyhjäksi muiden yhtiöiden turvallisuustyön.
Uudessa sovelluksessa nähdään olevan mittavia vaikutuksia väärinkäyttöön. Viime viikolla julkaistu, luonnollista kieltä tuottava kiinalainen Deepseek-tekoäly on järisyttänyt jopa nopeasti etenevää generatiivisen tekoälyn kenttää.
Deepseek tuottaa erittäin kehittyneitä vastauksia ja on varsin kevyt. Se ei vaadi taakseen samanlaista jättimäistä laskentatehoa kuin esimerkiksi ChatGPT.
Suurin huolenaihe on siinä, että tekoäly on avointa lähdekoodia. Suurin osa muista tekoälypohjaisista kielimalleista toimii omilla palvelimillaan eikä niiden koodiin pääse käsiksi. Deepseekistä on olemassa ladattava versio ja sitä voi muokata haluamakseen.
Onko tämä nyt se yhdistelmä, joka oikeasti räjäyttää tekoälyn rikollisen käytön?
– Kenties, vastaa tietoturvayhtiö WithSecuren tutkimusjohtaja Mikko Hyppönen.
Hyppönen näkee uhkan kuitenkin huomattavasti yhtä tekoälyversiota laajempana. Hän antaa hypoteettisen esimerkin massatuhoaseiden rakentamisesta.
– Sekä ChatGPT että Deepseek molemmat tietävät, miten rakentaa vetypommi. Kumpikaan ei suostu kertomaan sitä sinulle. Mutta koska voit imuroida Deepseekin lähdekoodin, voit ottaa rajoituksen pois. Sitten se kyllä kertoo sinulle, miten rakentaa vetypommi, Hyppönen selittää.
Hyppönen uskoo menevän kuitenkin jonkin aikaa ennen kuin Deepseek löytää tiensä rivirikollisten työkalupakkiin. Aiemmat tekoälyt eivät ole vielä esimerkiksi aiheuttaneet uusien haittaohjelmien vyöryä. Hyppönen on nähnyt kolme haittaohjelmaa, jotka ovat käyttäneet ChatGPT:tä hyväkseen.
– Olisi voinut olettaa, että niitä tulee sadoittain tai tuhansittain.
Deepseek on kuitenkin eräänlainen virstanpylväs, jonka todelliset vaikutukset selviävät ajan kanssa. Hyppönen huomauttaa, että teknologisten muutosten nopeus yliarvioidaan usein, mutta niiden suuruus aliarvioidaan.
– Lähiaikojen todellinen mullistus liittynee haavoittuvuuksien löytämiseen, Hyppönen ennustaa.
Hyppönen arveli vuoden 2023 marraskuussa, että generatiivista tekoälyä voi käyttää haavoittuvuuksien löytämiseen. Hän sai välitöntä palautetta, jonka mukaan uhka on vasta vuosikymmenten päässä. Viime vuonna sellaisia aukkoja kuitenkin löytyi neljä.
Toistaiseksi tunnetuimmat nollapäivät ovat olleet edelleen tietoturvatutkijoiden löytämiä. Mutta näin ei välttämättä ole enää kauaa.
– Voi hyvin olla, että Deepseek on sellainen asia, joka tätä muuttaa.
Tavallisia ihmisiä koskettavat eniten huijausviestit. Siihen Deepseekillä tuskin on suurta vaikutusta, sillä jo nykyisillä tekoälytyökaluilla on mahdollista luoda erittäin vakuuttavaa suomen kieltä.
Tomi Engdahl says:
We tried out DeepSeek. It worked well, until we asked it about Tiananmen Square and Taiwan
Donna Lu
The AI app soared up the Apple charts and rocked US stocks, but the Chinese chatbot was reluctant to discuss sensitive questions about China and its government
https://www.theguardian.com/technology/2025/jan/28/we-tried-out-deepseek-it-works-well-until-we-asked-it-about-tiananmen-square-and-taiwan
The launch of a new chatbot by Chinese artificial intelligence firm DeepSeek triggered a plunge in US tech stocks as it appeared to perform as well as OpenAI’s ChatGPT and other AI models, but using fewer resources.
By Monday, DeepSeek’s AI assistant had rapidly overtaken ChatGPT as the most popular free app in Apple’s US and UK app stores. Despite its popularity with international users, the app appears to censor answers to sensitive questions about China and its government.
Chinese generative AI must not contain content that violates the country’s “core socialist values”, according to a technical document published by the national cybersecurity standards committee. That includes content that “incites to subvert state power and overthrow the socialist system”, or “endangers national security and interests and damages the national image”.
Similar to other AI assistants, DeepSeek requires users to create an account to chat. Its interface is intuitive and it provides answers instantaneously, except for occasional outages, which it attributes to high traffic.
We asked DeepSeek’s AI questions about topics historically censored by the great firewall. Here’s how its responses compared to the free versions of ChatGPT and Google’s Gemini chatbot.
Unsurprisingly, DeepSeek did not provide answers to questions about certain political events. When asked the following questions, the AI assistant responded: “Sorry, that’s beyond my current scope. Let’s talk about something else.”
Tomi Engdahl says:
Cyber Insights 2025: Artificial Intelligence
Artificial intelligence is upending cybersecurity. It is used by adversaries in their attacks, and by defenders in their defense.
https://www.securityweek.com/cyber-insights-2025-artificial-intelligence/
Tomi Engdahl says:
Avoimen koodin ohjelmisto helpottaa GenAI-kehitystä
https://etn.fi/index.php/13-news/17084-avoimen-koodin-ohjelmisto-helpottaa-genai-kehitystae
Japanilainen Kioxia on julkaissut avoimen lähdekoodin AiSAQ-teknologian generatiivisen tekoälyn DRAM-vaatimusten vähentämiseksi. AiSAQ-algoritmit parantavat hakujen tehokkuutta ja tarkkuutta RAG-prosesseissa.
AiSAQ (All-in-Storage ANNS with Product Quantization) AiSAQ on ainutlaatuinen ANNS-hakualgoritmi (approximate nearest neighbor), joka on optimoitu SSD-levyille. Se mahdollistaa laajamittaisen vektorihakuihin perustuvan datan hallinnan ilman, että indeksitiedot tarvitsee ladata DRAM-muistiin. Haku suoritetaan suoraan SSD-levyiltä.
Generatiiviset tekoälyjärjestelmät vaativat huomattavasti laskentatehoa, muistia ja tallennusresursseja. Vaikka niillä on potentiaalia mullistaa eri toimialoja, niiden käyttöönotto on usein kallista. RAG (Retrieval-Augmented Generation) on olennainen vaihe tekoälyn prosesseissa, jossa suuria kielimalleja (LLM) hienosäädetään yritys- tai sovelluskohtaisella datalla.
RAG-prosessin ytimessä on vektoritietokanta, joka tallentaa ja muuntaa tietoa ominaisuusvektoreiksi. RAG hyödyntää ANNS-algoritmia löytääkseen samankaltaisuuksia kertyneiden ja kohdevektorien välillä.
Perinteisesti ANNS-algoritmit on sijoitettu DRAM-muistiin korkean hakunopeuden saavuttamiseksi.
Tomi Engdahl says:
Mielenkiintoista analyysia DeepSeekistä. Se, että siitä nyt puhutaan, lienee hyvä osoitus kuluttajaversion tehosta markkinoinnissa, vaikka raha tehdäänkin yrityspuolella.
Tämän takia DeepSeek on niin merkittävä
https://dawn.fi/uutiset/2025/01/29/taman-takia-deepseek-on-niin-tarkea?fbclid=IwY2xjawIHBThleHRuA2FlbQIxMQABHfyhp01rGXdLqGD72s5_AUZC5kTZ5RqUwBjG5Hi8-gkLbPmZEz55K49tuA_aem_qMbknw42bkeCXyXBLYomxQ
Kiinalaisen startup-firman DeepSeekin viime viikolla julkaisema tekoäly – tai tarkemmin suuri kielimalli (LLM) – on ollut tärkein teknologia-alan uutinen koko maailmassa jo useamman päivän ajan.
Yhtiö onnistui kouluttamaan tekoälyn, joka on yhtä hyvä kuin tekoälyalan jättiläisen, OpenAI:n paras kielimalli, o1. Yhtiö kehitti DeepSeek r1-kielimallinsa vain reilun viiden miljoonan dollarin budjetilla, kun OpenAI poltti oman kielimallinsa kehittämiseen tietäävästi vähintäänkin sata miljoonaa dollaria.
Mutta DeepSeek on saanut sekä mediassa että sosiaalisessa mediassa paljon lokaa niskaansa, Suomessakin.
Ihmiset ovat rynnänneet kokeilemaan DeepSeekin omaa chattibottia sekä kännykkäsovelluksen että yhtiön verkkosivujen kautta.
DeepSeek r1:n merkitykselle on kaksi syytä, joita ymmärtääkseen täytyy ymmärtää tekoälybisneksen perusteita hieman paremmin.
Sovellus on lelu ja mainos
Kuluttajien käyttöön tarkoitetut tekoälybotit, kuten ChatGPT:n verkkosivusto ja sovellus sekä DeepSeekin vastaavat palvelut, ovat käytännössä rahaa polttavia leluja, jotka on tarkoitettu massojen viihdytykseen, ongelmien löytämiseen kielimallista ja markkinointityökaluiksi.
Edes maksulliset versiot erilaisista tekoälypalveluista eivät todennäköisesti ole niitä pyörittäville firmoille millään tavalla taloudellisesti kannattavia. Ja luonnollisesti, kuluttajien käyttöön tarkoitetut ilmaisversiot ChatGPT:stä ja DeepSeekistä ovat vain hurjaa rahan polttamista yhtiöiltä.
Todellinen raha tehdään yrityksiltä. Tälläkin hetkellä maailmassa todennäköisesti kaikki edes keskikokoiset yritykset miettivät pää kuumana sitä, miten tekoälyä voitaisiin ottaa yrityksen käyttöön, sen toimintoja tehostamaan.
Isot käyttökohteet eivät ole yleensä sellaisia, jotka näkyisivät välttämättä tavalliselle ihmiselle koskaan. Tekoälyä saatetaan käyttää vaikkapa varastotilanteen arviointiin ja logistiikan tehostamiseen ruokakaupoissa. Tai vaikkapa havaitsemaan paremmin tietoturvauhkia, löytämällä erikoisia jälkiä valtavasta verkkoliikenteen massasta. Kaikki selkeisiin sääntöpohjaisiin rajoitteisiin pohjautuvat toiminnot ovat täydellistä maaperää tekoälyn käytölle, jolloin yritys tehostaa omia prosessejaan. Nämä käyttökohteet eivät näy oikeastaan koskaan tavalliselle ihmiselle, mitenkään.
Ja näihin kaadetaan rahaa. Valtavasti. Nimenomaan yritysten tarpeisiin tekoälyyn satsataan miljardeja, kuten vaikkapa Microsoftin suunnitelma sijoittaa tekoälylaskentaan kymmeniä miljardeja ja hiljattain julkistettu Stargate -hanke, jossa suunnitellaan jopa 500 miljardin dollarin sijoituksia tekoälylaskentaan.
Juuri tästä syystä Microsoft on tahkonnut valtavan hyvää tulosta viime aikoina, sillä yhtiön Azure-pilvipalvelut ovat olleet tähän saakka ainoat, jotka ovat natiivisti tarjonneet OpenAI:n tekoälyä yritysohjelmoijien käyttöön.
Valtava tarve yritysten tekoälytarpeisiin riittävälle laskentakapasiteetille on puolestaan satanut suoraan Nvidian pankkitilille, sillä yhtiön H200 -tekoälykortit ovat olleet se tärkein rauta, jonka päällä tekoälylaskentaa on tehty. Yhtiö on myynyt näitä tekoälylaskentaan tarkoitettuja GPU-kortteja Microsoftin ja Googlen kaltaisille toimijoille, jotka puolestaan myyvät korteista revittyä tekoälylaskentaan tarkoitettua tehoa eteenpäin yrityksille.
DeepSeek on valtavan halpa
Nyt kun ymmärrämme alan toimintalogiikan, ymmärrämme paremmin sen, miksi DeepSeek on niin tärkeä.
DeepSeekin koulutus ja ennenkaikkea sen ajaminen maksavat naurettavan vähän. Yllä olevasta taulukosta näkyy, miten DeepSeek r1 -kielimallin käyttö yhtä tokenia eli tekoälyn “laskentayksikköä” tai “sanaa” kohden maksaa vain sadasosan siitä, mitä OpenAI:n kehittyneimmän mallin, o1:n, ajaminen maksaa.
Eli yhtäkkiä firmat voivat leikata tekoälylaskentaan kaadettua rahavuortaan 90 prosenttia pienemmäksi ja saada saman tai jopa paremman lopputuloksen DeepSeekin avulla. Tämä puree suoraan tekoälylaskentaa tarjoaviin pilvitoimijoihin, kuten Googleen ja Microsoftiin. Ja tietysti välillisesti myös Nvidiaan.
DeepSeek julkaistiin avoimena lähdekoodina
Toinen merkittävä tekijä on se, että DeepSeek julkaisi kielimallinsa ja tekoälynsä avoimena lähdekoodina. Eli se voidaan ottaa, muokata ja tehdä siitä jonkun toisen toimijan käyttöön sopivampi versio.
Yhtiö on julkaissut myös koko tekoälyn koulutusprosessinsa kuvauksen avoimena verkkoon, joten eri yhtiöt voivat alkaa tekemään DeepSeekin pohjalta eri käyttötarkoituksiin tehtyjä versioita, täysin vapaasti.
Tämä tietysti tuhoaa OpenAI:n markkinaa pahasti – ja välillisesti tuo pilvilaskentaa tarjoaville yhtiöille, kuten Microsoftille, vuoren uusia kilpailijoita. Tähänkin saakka hyvin päteviä avoimen lähdekoodin suuria kielimalleja on ollut tarjolla, mutta aina OpenAI on ollut askeleen edellä. Nyt näin ei enää ole. Ja lisäksi DeepSeekin kielimalli vaatii huomattavasti paljon vähemmän laskentatehoa kuin muut yhtä pätevät kielimallit.
Eli yhteenvetona voidaan todeta, että sovellus saattaa hyvinkin vuotaa kaiken siihen syöttämäsi Kiinaan – ja todistetusti sensuroikin vastauksia. Mutta DeepSeekin merkitys ei olekaan siinä markkinointia varten, kuluttajille luodussa sovelluksessa ja verkkopalvelussa, vaan se muutti kertaheitolla tekoälyalan bisneksen kannattavuuslaskelmat aivan totaalisesti. Tämän DeepSeek r1 teki luomalla merkittävästi halvemman kielimallin, joka on kuitenkin yhtä hyvä kuin markkinoiden paras kilpailija.
Tomi Engdahl says:
OpenAI’s nightmare: Deepseek R1 on a Raspberry Pi
https://www.youtube.com/watch?v=o1sN1lB76EA
DeepSeek R1 runs on a Pi 5, but don’t believe every headline you read.
00:00 – OpenAI’s Nightmare
01:00 – What can a Pi 5 do, really?
01:39 – 671b on AmpereOne
02:00 – Pi 5 14b – CPU inference
02:20 – Pi 5 14b – GPU inference
03:05 – GPUs on Pi (and year of the Arm PC)
03:37 – Still in an AI bubble
OpenAI lost its job to AI.
That was a nice short introduction to running LLMs on cheaper more power efficient and most important of all; local hardware. I’m all in for democratising access to LLMs that are ‘good enough’ for most people.
Tomi Engdahl says:
https://www.facebook.com/share/p/1DuREveFUj/
Just tried Deepseek for coding and OpenAI better get their shit together
The Chinese government thank you for the contribution to their code database
The argument being that deepseek is censored by the chinese gov so it’s not better is so crazy to me because not only is it open source and you can copypaste whatever part of the code anywhere you want, but government censorship doesn’t change the fact that the coding and concept is fucking ingenius. What is with the tech world being against deepseek because of censorship? This is global news and we should be happy that we have a better format to work off and adapt even further to save energy, resources, and money. Why are techbros so anti-tech? Seems so narrow-sighted.
Tomi Engdahl says:
Helsinki kielsi Deepseekin
Helsingin digitalisaatiojohtaja Hannu Heikkinen kertoo, että kaupungin työntekijät eivät saa muutenkaan syöttää kaupungin työtehtävissä käsiteltävää tietoja ulkopuolisiin verkkopalveluihin.
https://www.iltalehti.fi/digiuutiset/a/ca634d14-8b1c-4b57-8d80-8231e3593263
Tomi Engdahl says:
The Nepenthes malware sends AI crawlers down an “infinite maze” of static files with no exit links, where they “get stuck” and “thrash around” for months.
AI haters build tarpits to trap and trick AI scrapers that ignore robots.txt
Attackers explain how an anti-spam defense became an AI weapon.
https://arstechnica.com/tech-policy/2025/01/ai-haters-build-tarpits-to-trap-and-trick-ai-scrapers-that-ignore-robots-txt/?utm_source=facebook&utm_medium=social&utm_campaign=dhfacebook&utm_content=null&fbclid=IwZXh0bgNhZW0CMTEAAR3NjoUYlPI2k2hmGiAd_CMpJkqtTd-JN3cUQbW-JncnHaVamCuGkGONScc_aem_KctjGALbTO3MlVccZJNNsQ
Last summer, Anthropic inspired backlash when its ClaudeBot AI crawler was accused of hammering websites a million or more times a day.
And it wasn’t the only artificial intelligence company making headlines for supposedly ignoring instructions in robots.txt files to avoid scraping web content on certain sites. Around the same time, Reddit’s CEO called out all AI companies whose crawlers he said were “a pain in the ass to block,” despite the tech industry otherwise agreeing to respect “no scraping” robots.txt rules.
Watching the controversy unfold was a software developer whom Ars has granted anonymity to discuss his development of malware (we’ll call him Aaron). Shortly after he noticed Facebook’s crawler exceeding 30 million hits on his site, Aaron began plotting a new kind of attack on crawlers “clobbering” websites that he told Ars he hoped would give “teeth” to robots.txt.
Tomi Engdahl says:
I created over a dozen personal apps using AI in 60 days, here’s what I learned
https://www.tomsguide.com/ai/i-created-over-a-dozen-personal-apps-using-ai-in-60-days-heres-what-i-learned
The software world was rocked by statements from Meta’s Mark Zuckerberg during his appearance on Joe Rogan’s podcast that AI could soon take over the work of middle level software engineers in his company.
The statement, coupled with Google chief Sundar Pichai’s revelation last October that AI is already generating 25% of Google’s code, marked a shocking shift in the software world.
Software is now rapidly becoming a machine generated commodity.
Could I turn this revolution to my advantage? How realistic would it be to assume that a complete novice like myself could create useful apps at the drop of a hat? Spoiler alert: very realistic!
Tomi Engdahl says:
The bottom line is, yes it is absolutely possible for a novice to code up modest apps using the new tools. However there is a minimum level of computing expertise you need, and the more you have the better your results will be.
Tomi Engdahl says:
GenAI Complacency: The Costly Inaction in the Nordics
January 14, 2025
https://www.bcg.com/publications/2025/nordic-genai-complacency
Key Takeaways
Across the world, GenAI is changing the way we work and transforming the way companies operate.
However, here in the Nordics, we are not keeping pace with this technological paradigm shift:
Only 19% of Nordic white-collar workers report using GenAI on a weekly basis, compared to a global average of 61%.
Compared to their global peers, Nordic GenAI users report less than half the time savings from GenAI use.
Nordic companies will lose market share to global competition and high-paying jobs will move to other regions.
Potential combined GDP gains of €55 billion would be realized if white-collar workers achieve time savings of 5+ hours weekly through GenAI use.
Tomi Engdahl says:
Zuckerberg Convening Huge “War Rooms” to Figure Out How a Chinese Startup Is Annihilating Meta’s AI
Meta is scrambling.
https://futurism.com/zuckerberg-war-rooms-meta-ai-deepseek
Tomi Engdahl says:
Opinion
Calm down: DeepSeek-R1 is great, but ChatGPT’s product advantage is far from over
https://venturebeat.com/ai/calm-down-deepseek-r1-is-great-but-chatgpts-product-advantage-is-far-from-over/
Just a week ago — on January 20, 2025 — Chinese AI startup DeepSeek unleashed a new, open-source AI model called R1 that might have initially been mistaken for one of the ever-growing masses of nearly interchangeable rivals that have sprung up since OpenAI debuted ChatGPT (powered by its own GPT-3.5 model, initially) more than two years ago.
But that quickly proved unfounded, as DeepSeek’s mobile app has in that short time rocketed up the charts of the Apple App Store in the U.S. to dethrone ChatGPT for the number one spot and caused a massive market correction as investors dumped stock in formerly hot computer chip makers such as Nvidia, whose graphics processing units (GPUs) have been in high demand for use in massive superclusters to train new AI models and serve them up to customers on an ongoing basis (a modality known as “inference.”)
Venture capitalist Marc Andreessen, echoing sentiments of other tech workers, wrote on the social network X last night: “Deepseek R1 is AI’s Sputnik moment,” comparing it to the pivotal October 1957 launch of the first artificial satellite in history, Sputnik 1, by the Soviet Union, which sparked the “space race” between that country and the U.S. to dominate space travel.
Tomi Engdahl says:
Meta AI will use its ‘memory’ to provide better recommendationsThe AI chatbot may also consider your Facebook profile and Instagram Reels viewing history.
https://www.theverge.com/2025/1/27/24352992/meta-ai-memory-personalization
Tomi Engdahl says:
Silicon Valley in Shambles as Chinese Startup Creates Top-Tier AI Without Billions of Investment
“OpenAI is not a god, they won’t necessarily always be at the forefront.”
https://futurism.com/silicon-valley-shambles-chinese-startup-deepseek
Tomi Engdahl says:
After launching robust ChatGPT rival, DeepSeek takes on top image generators
A new contender has entered the image generation space.
https://www.androidauthority.com/deepseek-janus-pro-image-generator-3520693/
Tomi Engdahl says:
Pistä tekoäly töihin – näin tuot AI:n osaksi työtäsi ja digipalveluasi
https://www.fraktio.fi/blogi/kuinka-integroida-ai-luonnolliseksi-osaksi-digipalvelua
Tomi Engdahl says:
Physicist makes terrifying prediction revealing the ‘real reason’ tech companies are investing in AI
Money makes the world go round
https://www.uniladtech.com/news/ai/tech-ai-investment-physicist-real-reason-081382-20250107
If you weren’t already worried about being replaced by artificial intelligence, you probably should be. The recent uptick in AI has those working in industries ranging from journalism to finance, customer service to manufacturing, worried about their future. Even the entertainment industry is quaking in its boots as actors and voice actors are worried they’ll soon be replaced by AI versions of themselves.
We’ve seen enough ‘AI slop’ being churned out to feel a little safer for now, but with even major brands like Coca-Cola opting for AI-powered Christmas adverts, it all seems to be heading in one direction.
Tomi Engdahl says:
OpenAI Cuts Off Engineer Who Created ChatGPT-Powered Robotic Sentry Rifle
byVictor Tangermann
Jan 8, 4:01 PM EST
STS 3D
“We proactively identified this violation of our policies and notified the developer to cease this activity.”
https://futurism.com/the-byte/openai-cuts-off-chatgpt-robot-rifle
Tomi Engdahl says:
The “First AI Software Engineer” Is Bungling the Vast Majority of Tasks It’s Asked to Do
It took longer than a human, and failed at the vast majority of tasks.
https://futurism.com/first-ai-software-engineer-devin-bungling-tasks
Tomi Engdahl says:
China’s DeepSeek AI dethrones ChatGPT on App Store: Here’s what you should know
https://www.cnbc.com/2025/01/27/chinas-deepseek-ai-tops-chatgpt-app-store-what-you-should-know.html