Artificial intelligence is rapidly changing many aspects of how we work and live. (How many stories did you read last week about self-driving cars and job-stealing robots? Perhaps your holiday shopping involved some AI algorithms, as well.) But despite the constant flow of news, many misconceptions about AI remain.
AI doesn’t think in our sense of the word at all, Scriffignano explains. “In many ways, it’s not really intelligence. It’s regressive.”
IT leaders should make deliberate choices about what AI can and can’t do on its own. “You have to pay attention to giving AI autonomy intentionally and not by accident,”
5,941 Comments
Tomi Engdahl says:
https://www.zdnet.com/article/ibm-open-sources-its-granite-ai-models-and-they-mean-business/
Tomi Engdahl says:
https://www.tietoevry.com/en/campaigns/responsible-ai/
Tomi Engdahl says:
Well played Google! DeepMind shows off Project Astra watching the OpenAI ChatGPT Voice announcement
https://www.tomsguide.com/ai/google-gemini/well-played-google-deepmind-shows-off-project-astra-watching-the-openai-chatgpt-voice-announcement
Tomi Engdahl says:
ChatGPT voi jäädä kakkoseksi – ”edelläkävijät hyödyntävät jo”
Tivi16.5.202422:31|päivitetty17.5.202415:56TEKOÄLY
Yritysjärjestelmiin upotetut generatiiviset mallit ovat muuttumassa yrityksille yhä houkuttelevammiksi vaihtoehdoiksi tekoälyn hyödyntämiseen.
https://www.tivi.fi/uutiset/chatgpt-voi-jaada-kakkoseksi-edellakavijat-hyodyntavat-jo/bab3ce4c-f423-4d1e-a60b-8e33f5bec4b9
Koneoppimisessa on aina ollut hyötyä siitä, että ongelma-alue rajataan tarkasti. Tämä pätee myös generatiivisiin malleihin. Mitä tarkemmasta kohteesta on kyse, sitä paremmin mallit suoriutuvat tehtävistään. Juuri tähän perustuvat OpenAI:n GPT-4:n kaltaisten laajojen kielimallien varjosta nousevat pienet ja kohdennetut kielimallit.
Siinä missä GPT-4:n on huhuttu vieneen mallin sisäisten muuttujien määrän biljoonaluokkaan, pienissä kielimalleissa puhutaan paljon matalammista parametrimääristä. Siksi ne jäävät yleiskäyttöisyydessä reippaasti suurten sisarustensa kuten esimerkiksi ChatGPT:n jälkeen. Silti ne voivat tarjota riittävästi tehoa yritysten tarkasti räätälöityihin tehtäviin.
Esimerkiksi Silo AI:n toimitusjohtaja Peter Sarlinin mukaan pienet generatiiviset mallit voivat olla rajatuissa kohteissa hyvinkin suorituskykyisiä ja samalla myös tarkasti hallittavia vaihtoehtoja.
Sarlin korostaa sitä, ettei generatiivisten mallien tarvitse jäädä yrityskäytössä pelkkiin työntekijöiden tehoa nostaviin yleiskäyttöisiin chatbotteihin. Pikemminkin niistä on tulossa komponentteja, jotka integroidaan niin liiketoiminnan tukijärjestelmiin kuin suoraan yrityksen tuotteisiin.
Tomi Engdahl says:
I’m convinced NVIDIA’s CEO was right about coding being dead in the water as a career option after watching OpenAI’s GPT-4o coding demo
News
By Kevin Okemwa published May 16, 2024
Despite being in its early release stages, GPT-4o is seemingly great at writing and detecting errors in code.
https://www.windowscentral.com/software-apps/im-convinced-nvidias-ceo-was-right-about-coding-being-dead-in-the-water-as-a-career-option-after-watching-openais-gpt-4o-coding-demo
Tomi Engdahl says:
Nvidia Blackwell ‘Superchips’ Will Cost Around $70,000 Each: Analyst
The company’s flagship, 72-GPU configuration will reportedly sell for around $3 million.
https://www.extremetech.com/computing/nvidia-blackwell-superchip-will-cost-around-70000-each-analyst
Tomi Engdahl says:
https://tekoalyolli.substack.com/
Tomi Engdahl says:
https://www.tomsguide.com/ai/chatgpt/chatgpt-with-gpt-4o-i-cannot-remember-the-last-time-i-was-this-blown-away-by-a-piece-of-technology
Tomi Engdahl says:
Google unleashes AI in search, raising hopes for better results and fears about less web traffic
https://techxplore.com/news/2024-05-big-google-ai-generated-results.html#google_vignette
Tomi Engdahl says:
OpenAI CEO Sam Altman’s “magical” GPT-4o felt more like routine Microsoft Copilot updates paired with a snub for Windows
Features
By Kevin Okemwa published May 14, 2024
OpenAI’s ChatGPT demos during the Spring update event were impressive, but “magical” might be a bit of a stretch.
https://www.windowscentral.com/software-apps/openai-ceo-magical-gpt-4o-paired-with-snub-for-windows
Tomi Engdahl says:
Just believing that an AI is helping boosts your performance
Minna Tiainen
Published: 13.5.2024
People perform better if they think they have an AI assistant – even when they’ve been told it’s unreliable and won’t help them
https://www.aalto.fi/en/news/just-believing-that-an-ai-is-helping-boosts-your-performance
Tomi Engdahl says:
FastGen: Cutting GPU Memory Costs Without Compromising on LLM Quality
https://www.marktechpost.com/2024/05/12/fastgen-cutting-gpu-memory-costs-without-compromising-on-llm-quality/
Tomi Engdahl says:
https://www.tomsguide.com/ai/chatgpt/openai-confirms-chatgpt-event-for-monday-feels-like-magic
Tomi Engdahl says:
Suomen suosituimmat tekoälykäyttötapaukset – Codenton tuore raportti perustuen 100 työpajaan
https://codento.com/fi/uutiset/suomen-suosituimmat-tekoalykayttotapaukset-codenton-tuore-raportti-perustuen-100-tyopajaan/
Tomi Engdahl says:
AI systems are already skilled at deceiving and manipulating humans, study shows
https://techxplore.com/news/2024-05-ai-skilled-humans.html#google_vignette
Tomi Engdahl says:
A framework to detect hallucinations in the text generated by LLMs
https://techxplore.com/news/2024-05-framework-hallucinations-text-generated-llms.html
Tomi Engdahl says:
https://www.hurja.fi/blogi/github-copilot-tekoalyn-hyodyntaminen-ohjelmistokehityksessa/
Tomi Engdahl says:
https://www.pcgamer.com/software/ai/openai-a-company-built-on-scraping-content-without-permission-makes-a-copyright-claim-against-a-subreddit-using-its-logo/
Tomi Engdahl says:
https://www.tomshardware.com/tech-industry/artificial-intelligence/commodore-64-can-use-ai-to-generate-8×8-sprites-takes-20-minutes-for-90-iterations
Tomi Engdahl says:
Reuters:
EU countries vote to approve the landmark EU AI Act, agreed in December 2023, to ensure “trust, transparency, and accountability”, taking effect in June 2024
Europe sets benchmark for rest of the world with landmark AI laws
https://www.reuters.com/world/europe/eu-countries-back-landmark-artificial-intelligence-rules-2024-05-21/
Tomi Engdahl says:
Kris Holt / Engadget:
Microsoft unveils Copilot+ PCs as a new class of AI-capable Windows PCs, offering at least 40 TOPs of NPU performance, with several OEMs and chipmakers on board — The aim is to handle more AI processing locally rather than in the cloud. — We’ve been hearing rumblings for months …
Microsoft unveils Copilot+ PCs with generative AI capabilities baked in
The aim is to handle more AI processing locally rather than in the cloud.
https://www.engadget.com/microsoft-unveils-copilot-pcs-with-generative-ai-capabilities-baked-in-170445370.html
Umar Shakir / The Verge:
Microsoft unveils Recall for Windows 11, an AI-powered search tool for Copilot+ PCs that keeps track of users’ actions and gives them an explorable timeline — Microsoft’s launching Recall for Windows 11, a new tool that keeps track of everything you see and do on your computer and …
We’ve been hearing rumblings for months now that Microsoft was working on so-called “AI PCs.” At a pre-Build event, the company spelled out its vision.
Microsoft is calling its version Copilot+ PCs, which CEO Satya Nadella described as a “new class of Windows PCs.” These contain hardware designed to handle more generative AI Copilot processes locally, rather than relying on the cloud. Doing so requires a chipset with a neural processing unit (NPU), and manufacturers such as Qualcomm have been laying the groundwork with chips like the Snapdragon X Elite.
Microsoft is taking a partner-first approach to making Copilot+ PCs. Along with chipmakers like AMD, Intel and Qualcomm, major OEMs including Acer, ASUS, Dell, HP and Lenovo are on board. The first Copilot+ laptops are available to preorder today and they’ll ship on June 18. Prices start at $999.
Tomi Engdahl says:
Umar Shakir / The Verge:
Microsoft unveils Recall for Windows 11, an AI-powered search tool for Copilot+ PCs that keeps track of users’ actions and gives them an explorable timeline — Microsoft’s launching Recall for Windows 11, a new tool that keeps track of everything you see and do on your computer and …
Recall is Microsoft’s key to unlocking the future of PCs
/ Microsoft’s groundbreaking new AI-powered tool can search and recall anything you’ve seen or done on your PC like a personal historian.
https://www.theverge.com/2024/5/20/24159258/microsoft-recall-ai-explorer-windows-11-surface-event
Tomi Engdahl says:
Joanna Stern / Wall Street Journal:
Impressions of Windows 11 AI features Recall, Live Captions, and Cocreator, and a chat with Satya Nadella on how the Copilot+ PC push will outperform MacBooks — Microsoft CEO Satya Nadella told our columnist new efficient, powerful chips equip Windows Copilot+ laptops to compete with Apple’s MacBooks
https://www.wsj.com/tech/personal-tech/microsoft-ceo-satya-nadella-interview-ai-laptops-76eef1e1?st=a5ugbgvcuzb1clf&reflink=desktopwebshare_permalink
Tomi Engdahl says:
Lawrence Bonk / Engadget:
Acer, Asus, Dell, HP, Samsung, and Lenovo joined Microsoft in debuting laptops featuring Qualcomm’s Snapdragon X Elite or Plus chips and a dedicated Copilot key — They all have similar processors, but there are plenty of differences in design and functionality.
Here are all of the just-announced Copilot+ PCs with Snapdragon X Chips
They all have similar processors, but there are plenty of differences in design and functionality.
https://www.engadget.com/here-are-all-of-the-just-announced-copilot-pcs-with-snapdragon-x-chips-184825090.html
We knew more computers were coming that would feature a native version of Microsoft’s AI Copilot toolset, but we didn’t quite know how many were set to be announced. It’s practically an AI avalanche. Companies like Dell, Acer and HP have all just announced computers that have adopted Microsoft’s AI software and NVIDIA’s AI hardware. The age of the AI PC is upon us.
Tomi Engdahl says:
Tom Warren / The Verge:
Microsoft’s new push for Windows PCs with Arm chips promises better emulation for non-native apps and Qualcomm chips that can best an M3 in a MacBook Air
Inside Microsoft’s mission to take down the MacBook Air
/ Microsoft is convinced its new Copilot Plus PCs will beat Apple’s M3 processor and spark a new generation of Windows laptops.
https://www.theverge.com/2024/5/20/24160463/microsoft-windows-laptops-copilot-arm-chips-m1
Tomi Engdahl says:
Jeremy Kahn / Fortune:
A look at Satya Nadella’s approach to pushing AI at Microsoft, as the company bets big on OpenAI and in-house models, but also on G42, Mistral, Figure, and more
Satya Nadella has made Microsoft 10 times more valuable in his decade as CEO. Can he stay ahead in the AI age?
https://fortune.com/2024/05/20/satya-nadella-microsoft-build-generative-ai-openai/
Something has caught Satya Nadella’s attention. It’s a small thing—just a five-letter word inside a box, lurking in the corner of a complicated PowerPoint slide, flashed on a screen for a fraction of a second inside a convention hall in Kuala Lumpur, Malaysia. But it niggles. “You’re doing also Llama? You’re using both?” Nadella asks, a note of surprise in his voice.
Llama is the AI model, not the animal. It is a piece of open-source software created by Meta, the social media giant that has pivoted hard to AI and is competing with Microsoft and others to dominate the foundations of the emerging generative AI economy. “Both” is a reference to the fact that this Malaysian agriculture technology company—chosen to show off its use of Microsoft’s technology to Microsoft’s CEO— is using Meta’s rival AI model in addition to GPT-4, the large language model (LLM) created by Microsoft’s strategic partner, OpenAI. Nadella wants his Redmond, Wash.–based software giant to have the most capable, popular AI models on the market.
Tomi Engdahl says:
Bloomberg:
OpenAI pulls ChatGPT’s Sky voice after users compared the sound to Scarlett Johansson, saying Sky is an actress’ voice and wasn’t chosen to be an “imitation” — – Selecting the Sky profile redirects ChatGPT users to Juniper — Company said it was not trying to imitate Johansson’s voice
OpenAI Pulls Johansson Soundalike Sky’s Voice From ChatGPT
Selecting the Sky profile redirects ChatGPT users to Juniper
Company said it was not trying to imitate Johansson’s voice
https://www.bloomberg.com/news/articles/2024-05-20/openai-to-pull-johansson-soundalike-sky-s-voice-from-chatgpt
Tomi Engdahl says:
Future AI trends: the move from generative to interactive AI
https://www.ft.com/partnercontent/infobip/future-ai-trends-the-move-from-generative-to-interactive-ai.html?utm_source=FB&utm_medium=technology&utm_content=paid&fbclid=IwAR23duFE4uLVprEyTvoziceWoYMKfaRuhdQC5X6c8Bzgo6UlG6ZUQ5la4ew_aem_ARp9uW0MQ6Rdh-nwECKALUUTTG7EHWOXvNokYk3V1O3-uM_y2weMWO2nUXwrDyOiGEHRsiVROzj-pyOUNYvRslNV&utm_id=120206853243350278&utm_term=120209072833990278&utm_campaign=120206853243350278
Generative AI has helped to transform the customer experience. Interactive AI will revolutionise it.
Over the past year, the rise and evolution of generative AI has transformed the customer experience. End-to-end conversational customer journeys are quickly becoming the norm for businesses and brands.
Interactive AI will help the customer experience become richer, more intuitive, and more profitable for businesses. Brands must better understand these changes and how they can implement the latest technology to delight their customers and deliver the experience consumers will come to expect.
For instance, Infobip’s latest messaging trends report shows businesses and customers increasingly prefer conversational messaging channels. In 2023, Infobip recorded a 137% increase in mobile app messages compared to 2022, a 73% rise in social media messages and a 63% increase in chat app messages.
Conversational AI, powered by generative AI, has been a driving force behind these trends. Advanced AI, which uses machine learning and natural language processing to replicate human conversation by recognising words, intent, and context, enables businesses to deliver flexible experiences, have a personality and know when to hand over to a human agent.
Tomi Engdahl says:
generated code — NetBSD and Gentoo lead the charge on forbidding AI-written code
News
By Christopher Harper published 3 days ago
Not all FOSS (Free and Open Source Software) developers want AI messing with their code.
https://www.tomshardware.com/software/linux/linux-distros-ban-tainted-ai-generated-code
Tomi Engdahl says:
How Microsoft Copilot and ServiceNow Now Assist enhance employee and IT admin choice and flexibility
https://partner.microsoft.com/fi-FI/blog/article/how-microsoft-copilot-and-servicenow-now-assist-enhance-choice-and-flexibility
Tomi Engdahl says:
https://www.marktechpost.com/2024/05/10/this-ai-paper-by-microsoft-and-tsinghua-university-introduces-yoco-a-decoder-decoder-architectures-for-language-models/
Tomi Engdahl says:
https://www.karhuhelsinki.fi/blogi/google-ai-overview-tekoaly-tuottaa-hakutuloksia/
Tomi Engdahl says:
AI Companies Make Fresh Safety Promise at Seoul Summit, Nations Agree to Align Work on Risks
Leading artificial intelligence companies made pledge to develop AI safely, while world leaders agreed to build a network of publicly backed safety institutes to advance research and testing of the technology.
https://www.securityweek.com/ai-companies-make-fresh-safety-promise-at-seoul-summit-nations-agree-to-align-work-on-risks/
Tomi Engdahl says:
Tom Warren / The Verge:
Microsoft will soon let businesses build custom Copilot AI agents to automate tasks, and unveils Team Copilot to help with tasks in Teams, Loop, and Planner — Microsoft will soon allow businesses and developers to build AI-powered Copilots that can work like virtual employees and perform tasks automatically.
Microsoft’s new Copilot AI agents act like virtual employees to automate tasks
/ Microsoft argues that its AI automation will remove the boring bits of jobs instead of replacing jobs entirely.
https://www.theverge.com/2024/5/21/24158030/microsoft-copilot-ai-automation-agents
Tomi Engdahl says:
Belle Lin / Wall Street Journal:
How Pinecone and other startups use vector databases and retrieval-augmented generation, or RAG, to help businesses link their private data with LLMs
How a Decades-Old Technology and a Paper From Meta Created an AI Industry Standard
Vector databases were an obscure technology until Pinecone and other startups turned it into a billion-dollar market amid the AI boom
https://www.wsj.com/articles/how-a-decades-old-technology-and-a-paper-from-meta-created-an-ai-industry-standard-354a810e?st=g92jj70ev26pep4&reflink=desktopwebshare_permalink
Businesses are using a relatively unknown technology called vector databases to help solve the problem of using their private data with generative artificial intelligence. And it’s turning the tool, and its makers, into some of the AI boom’s biggest winners.
Vector databases have been around for decades, but are now emerging as something of an industry standard for AI businesses to use alongside a technique called retrieval-augmented generation, or RAG.
When combined, businesses can link their private data with large-language models like OpenAI’s GPT-4, allowing the AI to perform data analysis, summarization and other tasks on their data. Without them, AI models are limited to what they have learned from their initial training on public data online, up to a certain point in time, and are more prone to factual errors called “hallucinations.”
“We hit the inflection point with ChatGPT and OpenAI and all the tools that exploded with it,” said Edo Liberty, Pinecone’s founder and chief executive. “This happened just as we had matured our platform, and we were well ahead of the competition.”
New arrivals
Gartner analyst Arun Chandrasekaran said Pinecone’s success is tied to its early entry in the market, that it manages the product for customers in the cloud, and its integration with AI models and other developer tools.
But Pinecone is no longer alone.
At least eight vector-database startups have entered the market since 2022, according to data provider Crunchbase. Investors have put nearly $1 billion into vector-database startups in that time, $73 million of which has been invested so far this year across eight deals, Crunchbase found.
Existing database vendors such as Oracle, MongoDB and Redis also have released their own vector databases, and tech giants Amazon, Microsoft and Google have added similar capabilities to their products. There are also open-source contenders.
More may be coming. The global vector-database market is expected to grow to $4.3 billion by 2028, from $1.5 billion in 2023, according to Markets and Markets, a research firm in India.
While fewer than 2% of enterprises were using the technology in September 2023, that number will grow to over 30% in the next two years, Gartner predicts.
Vector databases are finding a new use by enterprise and AI developers because they power the RAG technique. Originating from a 2020 paper by an AI research group at Meta Platforms, RAG is commonly used by enterprises to build chatbots for employees to reference company policies, or for customer service and salespeople to pull information from knowledge bases.
Vector databases are different from traditional databases with columns and rows because they are designed to store a massive amount of data as “vectors,” or numerical representations of the raw data. That makes them ideal for RAG, the process where generative AI models pull from large amounts of vector data to improve their responses with the additional information.
RAG vs. fine-tuning
Compared with fine-tuning—a technique used to create a custom AI model based on an existing large language model—RAG is cheaper because it doesn’t require massive amounts of computing power and advanced AI expertise, and is easier and faster for developers to implement. Roughly 80% of enterprises are using RAG, compared with 20% using fine-tuning, Chandrasekaran said.
Cristina Pieretti, general manager of digital insights for Moody’s, said the rating firm’s financial intelligence group used RAG to connect its research and ratings data with a model from OpenAI. RAG helps the AI model provide up-to-date financial information when customers ask its research assistant to assess investments and compare entities.
Many enterprises are already using RAG whether they’re aware of it or not, Gartner’s Chandrasekaran said.
Microsoft’s Copilot for Security, which aims to help cybersecurity pros with tasks like incident management, relies on RAG to link the AI with customers’ own security data, said Vasu Jakkal, the company’s corporate vice president of security, compliance, identity, management and privacy.
Many of Pinecone’s clients are “AI builders”—the developers and companies making generative AI-based products—who use RAG with vector databases under the hood for their customers, Liberty said.
Tomi Engdahl says:
Ken Yeung / VentureBeat:
Microsoft partners with Khan Academy, moving the Khanmigo agent, used by 65K+ students, to its Azure OpenAI Service and making it free for US K-12 teachers
https://venturebeat.com/ai/microsoft-to-enhance-khan-academys-ai-bot-with-azure-openai-service-develop-phi-3-math-model/
Tomi Engdahl says:
Frederic Lardinois / TechCrunch:
Microsoft launches Copilot Extension for GitHub, letting developers build third-party skills into Copilot, starting with DataStax, Stripe, MongoDB, and more
GitHub Copilot gets extensions
https://techcrunch.com/2024/05/21/github-copilot-gets-extensions/
Tomi Engdahl says:
Emma Roth / The Verge:
Microsoft debuts a new AI feature for Edge to translate spoken content via dubbing and subtitles live, on YouTube, LinkedIn, Coursera, news sites, and more
Microsoft Edge will translate and dub YouTube videos as you’re watching them
/ It will also support AI dubbing and subtitles on LinkedIn, Coursera, Bloomberg, and more
https://www.theverge.com/2024/5/21/24160664/microsoft-edge-real-time-video-translation-ai
Tomi Engdahl says:
Ken Yeung / VentureBeat:
Microsoft announces the general availability of its Phi-3 models, including Phi-3-Silica, a 3.3B parameter model that will be embedded on all Copilot+ PCs — Microsoft is making more investments in the development of small language models. At its Build developer conference …
Microsoft introduces Phi-Silica, a 3.3B parameter model made for Copilot+ PC NPUs
https://venturebeat.com/ai/microsoft-introduces-phi-silica-a-3-3b-parameter-model-made-for-copilot-pc-npus/
Tomi Engdahl says:
Kevin Okemwa / Windows Central:
Microsoft ships Azure AI Studio in broad availability, adds support for OpenAI’s GPT-4o, and announces a new multimodal model in its lightweight Phi-3 family
https://www.windowscentral.com/software-apps/microsoft-brings-its-lightweight-phi-3-model-and-openais-magical-gpt-4o-to-azure-ai-to-help-devs-build-transformational-experiences
Tomi Engdahl says:
Imran Rahman-Jones / BBC:
The UK ICO is “making enquiries with Microsoft” over the company’s Recall feature, which some privacy campaigners have called a potential “privacy nightmare” — The UK data watchdog says it is “making enquiries with Microsoft
UK watchdog looking into Microsoft AI taking screenshots
https://www.bbc.com/news/articles/cpwwqp6nx14o
The UK data watchdog says it is “making enquiries with Microsoft” over a new feature that can take screenshots of your laptop every few seconds.
Microsoft says Recall, which will store encrypted snapshots locally on your computer, is exclusive to its forthcoming Copilot+ PCs.
But the Information Commissioner’s Office (ICO) says it is contacting Microsoft for more information on the safety of the product, which privacy campaigners have called a potential “privacy nightmare”.
Microsoft says Recall is an “optional experience” and it is committed to privacy and security.
According to its website, users “can limit which snapshots Recall collects”.
“Recall data is only stored locally and not accessed by Microsoft or anyone who does not have device access,” the firm said in a statement.
And it said a would-be hacker would need to gain physical access to your device, unlock it and sign in before they could access saved screenshots.
But an ICO spokesperson said firms must “rigorously assess and mitigate risks to peoples’ rights and freedoms” before bringing any new products to market.
“We are making enquiries with Microsoft to understand the safeguards in place to protect user privacy,” they said.
Tomi Engdahl says:
Louise Matsakis / Wired:
Google is rolling out a new AI-powered ad format that shows shoppers how items of clothing would look on different skin tones and body types — A new breed of online ad allows brands to pay Google to offer shoppers AI-generated images that show how items of clothing would look on different skin tones and body types.
Google Taps AI to Show Shoppers How Clothes Fit Different Bodies
A new breed of online ad allows brands to pay Google to offer shoppers AI-generated images that show how items of clothing would look on different skin tones and body types.
https://www.wired.com/story/google-ai-shopping-clothes-fit-different-bodies/
Tomi Engdahl says:
Charlie Warzel / The Atlantic:
OpenAI’s Scarlett Johansson debacle is merely a reminder of AI’s manifest destiny philosophy: “This is happening, whether you like it or not”, consent be damned — The Scarlett Johansson debacle is a microcosm of AI’s raw deal: It’s happening, and you can’t stop it.
OpenAI Just Gave Away the Entire Game
https://www.theatlantic.com/technology/archive/2024/05/openai-scarlett-johansson-sky/678446/?gift=2iIN4YrefPjuvZ5d2Kh306LFFT6yU6HZVP_tmIcOfig&utm_source=copy-link&utm_medium=social&utm_campaign=share
The Scarlett Johansson debacle is a microcosm of AI’s raw deal: It’s happening, and you can’t stop it.
By Charlie Warzel
Photo collage of Scarlett Johansson, the OpenAI logo, and Sam Altman
Illustration by Paul Spella / The Atlantic. Sources: Jason Redmond / AFP; Paul Morigi / Getty.
May 21, 2024, 5:54 PM ET
If you’re looking to understand the philosophy that underpins Silicon Valley’s latest gold rush, look no further than OpenAI’s Scarlett Johansson debacle. The story, according to Johansson’s lawyers, goes like this: Nine months ago, OpenAI CEO Sam Altman approached the actor with a request to license her voice for a new digital assistant; Johansson declined. She alleges that just two days before the company’s keynote event last week, in which that assistant was revealed as part of a new system called GPT-4o, Altman reached out to Johansson’s team, urging the actor to reconsider. Johansson and Altman allegedly never spoke, and Johansson allegedly never granted OpenAI permission to use her voice. Nevertheless, the company debuted Sky two days later—a program with a voice many believed was alarmingly similar to Johansson’s.
Johansson told NPR that she was “shocked, angered and in disbelief that Mr. Altman would pursue a voice that sounded so eerily similar to mine.” In response, Altman issued a statement denying that the company had cloned her voice and saying that it had already cast a different voice actor before reaching out to Johansson. (I’d encourage you to listen for yourself.) Curiously, Altman said that OpenAI would take down Sky’s voice from its platform “out of respect” for Johansson. This is a messy situation for OpenAI, complicated by Altman’s own social-media posts. On the day that OpenAI released ChatGPT’s assistant, Altman posted a cheeky, one-word statement on X: “Her”—a reference to the 2013 film of the same name, in which Johansson is the voice of an AI assistant that a man falls in love with. Altman’s post is reasonably damning, implying that Altman was aware, even proud, of the similarities between Sky’s voice and Johansson’s.
Tomi Engdahl says:
Steven Levy / Wired:
Anthropic researchers detail attempts to peer inside the “black box” of LLMs, learning which combinations of neurons evoke specific concepts — What goes on in artificial neural networks work is largely a mystery, even to their creators. But researchers from Anthropic have caught a glimpse.
AI Is a Black Box. Anthropic Figured Out a Way to Look Inside
What goes on in artificial neural networks work is largely a mystery, even to their creators. But researchers from Anthropic have caught a glimpse.
https://www.wired.com/story/anthropic-black-box-ai-research-neurons-features/
Tomi Engdahl says:
Michael Nuñez / VentureBeat:
Generative AI music startup Suno raised $125M from Lightspeed Venture Partners, Nat Friedman, Daniel Gross, Matrix, Founder Collective, and others — Suno, a trailblazing music AI startup, announced today that it has raised a staggering $125 million in its latest funding round …
Suno raises $125M to become the ‘ChatGPT’ of music creation
https://venturebeat.com/ai/suno-raises-125m-to-become-the-chatgpt-of-music-creation/
Tomi Engdahl says:
Joyce Lee / Reuters:
Google, Meta, Microsoft, OpenAI, Amazon, IBM, and 10 other companies commit to safe AI development at the AI Seoul Summit 2024, hosted by South Korea and the UK
Second global AI summit secures safety commitments from companies
https://www.reuters.com/technology/global-ai-summit-seoul-aims-forge-new-regulatory-agreements-2024-05-21/
Tomi Engdahl says:
Tim Bradshaw / Financial Times:
How Meta, Microsoft, and Google are helping OpenAI develop Triton, a tool to help run code efficiently on AI chips to compete with Nvidia’s CUDA platform
Nvidia’s rivals take aim at its software dominance
OpenAI-led push for an alternative to Cuda seeks to break chipmaker’s stranglehold on AI market
https://www.ft.com/content/320f35de-9a6c-4dbf-b42f-9cdaf35e45bb
Tomi Engdahl says:
Jess Weatherbed / The Verge:
Adobe adds generative AI tools to Lightroom, including Firefly-powered Generative Remove to let users “paint” over unwanted objects, available in early access
Adobe Lightroom gets a magic eraser, and it’s impressive
/ Generative Remove can delete objects from images and generate convincing backgrounds to patch the blank spaces.
https://www.theverge.com/2024/5/21/24161490/adobe-lightroom-generative-remove-object-removal-tool
Tomi Engdahl says:
https://hackaday.com/2024/05/15/a-slice-of-simulation-google-sheets-style/
Tomi Engdahl says:
https://hackaday.com/2024/05/15/how-ai-large-language-models-work-explained-without-math/