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,”
5,940 Comments
Tomi Engdahl says:
Intel, AMD take a back seat as Qualcomm takes center stage in Microsoft’s AI PC push
Plus: Windows set for ML-powered always-watching-you Recall feature
https://www.theregister.com/2024/05/21/qualcomm_windows_microsoft/
Tomi Engdahl says:
STUDENTS SHOW UP TO GRADUATION, FIND COMMENCEMENT SPEAKER IS AN AI ROBOT
https://futurism.com/the-byte/students-graduation-speaker-ai-robot
Tomi Engdahl says:
https://www.tieturi.fi/koulutusala/tekoaly-ai/
Tomi Engdahl says:
https://blog.netprofile.fi/tekoaly-tehostaa-tyota-mutta-miten-rakentaa-sen-avulla-aitoa-ja-eettisesti-vahvaa-viestintaa
Tomi Engdahl says:
https://dashboa.com/fi/sisallontuotanto-tekoalylla/
Tomi Engdahl says:
Juristin vinkit teknologian ja tekoälyn käyttöönottoon
https://fondia.com/fi/fi/ajankohtaista/artikkelit/juristin-vinkit-teknologian-ja-tekoalyn-kayttoonottoon
Organisaation tekoälymatkan huoneentauluksi voidaan nostaa kolme ydinohjetta:
Tunne asiaan liittyvä sääntely
Kommunikoi, kouluta, kannusta
Tiedä mitä hankit
Tomi Engdahl says:
https://xn--tekolysovellukset-tqb.fi/tekoaly-suomessa/
Tomi Engdahl says:
https://digia.com/blogi/digian-kyselytutkimus-paljasti-yllattavia-tietoja-ai-haasteista-tekoalyn-tiella-on-suomessa-merkittava-ongelma#amp_tf=L%C3%A4hde%3A%20%251%24s&aoh=17174805357952&referrer=https%3A%2F%2Fwww.google.com&share=https%3A%2F%2Fdigia.com%2Fblogi%2Fdigian-kyselytutkimus-paljasti-yllattavia-tietoja-ai-haasteista-tekoalyn-tiella-on-suomessa-merkittava-ongelma
Tomi Engdahl says:
The future of auditing is AI-powered
Tomi Engdahl says:
CEOS COULD EASILY BE REPLACED WITH AI, EXPERTS ARGUE
https://futurism.com/the-byte/ceos-easily-replaced-with-ai
CEOs better start endearing themselves to their employees real quick, because oh boy: the case for replacing them with AI just keeps mounting.
“Some people like the social aspects of having a human boss,” Phoebe Moore, professor of management and the futures of work at the University of Essex Business School, told The New York Times. “But after COVID, many are also fine with not having one.”
The idea of subbing bosses for bots is the flipside to all those fears about AI causing widespread job destruction, which often focus on the fact — quite justifiably — that it’ll be regular grunts being shown the door in favor of intelligent automation.
Tomi Engdahl says:
Muusikko kuuli, miten tekoäly tekee biisin kahdessa minuutissa – nyt hän pohtii alanvaihtoa
Tekoäly tulee vauhdilla musiikkialalle. Musiikintekijä Torsti Spoof varautuu töiden loppumiseen.
https://yle.fi/a/74-20089222
Tomi Engdahl says:
https://www.marktechpost.com/2024/06/01/this-ai-paper-from-princeton-and-the-university-of-warwick-proposes-a-novel-artificial-intelligence-approach-to-enhance-the-utility-of-llms-as-cognitive-models/
Tomi Engdahl says:
https://www.itewiki.fi/p/onko-generatiivinen-ai-ehtinyt-viela-rantautua-suomalaisyrityksiin
Tomi Engdahl says:
https://www.uusiteknologia.fi/2024/06/04/yritykset-avautuivat-tekoalykokemuksistaan/
Thompson nimesi tekoälyn suurimmaksi teknologiseksi innovaatioksi sitten internetin. Hänen mukaansa monen yrityksen tekoälymatkan hidasteena on kuitenkin data, joka ei ole valmista käytettäväksi tekoälyn polttoaineena.
“Riippuen keneltä kysyt, vain 4–34 prosenttia ammattilaisista uskoo, että heillä on käytössään tekoälylle sopivaa dataa”, Thomson kertoi. Hän ohjeistikin yleisöä olemaan syöttämättä kaikkea dataa tekoälylle ja aloittamaan sen sijaan datasta, joka on varmasti luotettavaa.
Salesforce Suomen maajohtaja Taira Tepponen kertoi, että tapahtuman tavoitteena oli “tuoda esiin konkretiaa ja tosielämän esimerkkejä siitä, miten suomalaisyritykset hyötyvät tekoälystä nyt ja lähitulevaisuudessa.”
Esimerkiksi VTT:n Mikael Back ja Timo Kakkonen nostivat esiin tilaisuudessa kulttuurin merkityksen. Backin mukaan osan tekoälytyökaluista voi omaksua vartissa, eli aina esteenä ei ole teknologian haastavuus. Sen sijaan yrityksissä pitäisi saada aikaan kulttuurin muutos, jotta tekoälyn hyödyt voidaan valjastaa käyttöön.
Tomi Engdahl says:
Korvaako tekoäly toimittajan?
https://etn.fi/index.php/opinion/16291-korvaako-tekoaely-toimittajan
Monissa lehtiartikkeleissa lukee jo, että tekoälyä on käytetty esimerkiksi laatimaan artikkelin lyhennelmä. Joskus näkee, että käännöksissä on ”käytetty apuna tekoälyä”. Pitäisikö siis toimittajien alkaa etsiä uusia hommia, kuten säveltäjien, jotka jo – ainakin Ylen artikkelin mukaan – joutuvat etsimään uusia töitä?
Tukholmassa järjestetyssä AWS Summitissa pääsin kokeilemaan GenAI:n käyttöä artikkelin otsikkojen generoinnissa. Käytännössä prosessissa valitaan malli (demossa Claude Sonnet 3), inferenssin eli päättelyn parametrit (satunnaisuus/diversiteetti), mallin noudattamista kuvaava lämpötila, tuotoksen pituus ja tämän jälkeen materiaalista tuotetaan promptilla haluttuja tuloksia.
Aineistona oli itse tehty artikkeli, joka syötettiin Bedrockiin. Keskeistä oli kuvaus siitä, kenen tekemiä otsikoita haettiin. ”Asiallinen journalisti” -moodissa otsikot eivät olleet kovin hyviä. Sen sijaan tabloid- eli iltapäivälehtimoodissa otsikot olivat parempia. Eivät ns. klikkiotsikoita, vaan yksinkertaisesti parempia.
Työpajassa Bedrockia testattiin myös artikkelin lyhennelmien tekoon. Tulokset olivat pääosin erinomaisia. Anthropicin Claude Sonnet 3 -malli ei tee virheitä, piste. Kun malli työskentelee vain annetun datan kanssa ja tiukasti määritellyillä parametreilla, se ei hallusinoi.
Lopputuloksena voi arvioida, että aivan vielä AI ei ole viemässä toimittajia kortistoon. Sitä voi kuitenkin käyttää monenlaiseen sparraamiseen ja aika moni käyttääkin jo.
Toistaiseksi tekoäly ei kuitenkaan pysty esimerkiksi tiedotteiden merestä valitsemaan tärkeimpiä tai relevanteimpia aiheita.
Samat uhat ja mahdollisuudet koskevat monia muitakin ammatteja, erityisesti ns. luovilla aloilla. Tekoäly pitäisikin mahdollisimman nopeasti ottaa oleelliseksi osaksi kaikkia ammattikoulutuksia
Tomi Engdahl says:
https://www.securityweek.com/former-openai-employees-lead-push-to-protect-whistleblowers-flagging-artificial-intelligence-risks/
Tomi Engdahl says:
Kevin Roose / New York Times:
Current and former OpenAI and Google DeepMind staff sign an open letter warning of a culture of recklessness and secrecy at “frontier AI companies” — Daniel Kokotajlo, a former researcher in OpenAI’s governance division, is an organizer of a group of former and current employees …
https://www.nytimes.com/2024/06/04/technology/openai-culture-whistleblowers.html?u2g=i&unlocked_article_code=1.xE0.qXLu.0XwdynmvKcft&smid=url-share
OpenAI Insiders Warn of a ‘Reckless’ Race for Dominance
A group of current and former employees is calling for sweeping changes to the artificial intelligence industry, including greater transparency and protections for whistle-blowers.
Tomi Engdahl says:
Shakeel Hashim / Transformer:
Former OpenAI researcher Leopold Aschenbrenner says he was fired in April for sharing a memo raising concerns about OpenAI’s security practices with the board — Leopold Aschenbrenner also said he was interrogated about his team’s “loyalty to the company” — Leopold Aschenbrenner
OpenAI employee says he was fired for raising security concerns to board
Leopold Aschenbrenner also said he was interrogated about his team’s “loyalty to the company”
https://www.transformernews.ai/p/openai-employee-says-he-was-fired
Leopold Aschenbrenner, an OpenAI safety researcher who was reportedly fired earlier this year for “leaking”, has now offered his side of the story.
In an interview with Dwarkesh Podcast, published on Tuesday, Aschenbrenner said OpenAI told him he was fired for raising security concerns to the company’s board. Aschenbrenner was also ousted for sharing a document that OpenAI alleged contained sensitive information, a charge which he denies.
Aschenbrenner said he first received a formal warning from OpenAI’s HR department after he raised concerns about OpenAI’s security practices to the board. “I wrote an internal memo about OpenAI’s security, which I thought was egregiously insufficient to protect against the theft of model weights or key algorithmic secrets from foreign actors,” Aschenbrenner said, noting that he “shared this memo with a few colleagues and a couple of members of leadership, who mostly said it was helpful.”
A few weeks later, OpenAI suffered a major security incident, Aschenbrenner said, prompting him “to share the memo with a couple of board members.” But he was swiftly reprimanded. “It was made very clear to me that leadership was very unhappy I had shared this memo with the board,” he said. “Apparently, the board hassled leadership about security.”
Aschenbrenner said he received an “official HR warning” for this. Though he wasn’t fired immediately, when he was fired several months later the company told him this incident was a contributing factor. “When I was fired, it was made very explicit that the security memo was a major reason for my being fired,” Aschenbrenner said. “They said, ‘the reason this is a firing and not a warning is because of the security memo.’”
Tomi Engdahl says:
Fanny Potkin / Reuters:
Sources: some Chinese AI chip companies, including MetaX and Enflame, are submitting less powerful GPU designs to TSMC to comply with US sanctions
Exclusive: Chinese AI chip firms downgrading designs to secure TSMC production
https://www.reuters.com/technology/chinese-ai-chip-firms-downgrading-designs-secure-tsmc-production-sources-say-2024-06-04/
Tomi Engdahl says:
Alex Konrad / Forbes:
How billionaire investors of the internet era are fighting to determine whether AI’s future will be one of concentrated safety, or of unfettered advancement — Billionaire investors of the internet era are now locked in a war of words and influence to determine whether AI’s future …
Vinod Khosla, Marc Andreessen And The Billionaire Battle For AI’s Future
https://www.forbes.com/sites/alexkonrad/2024/06/04/inside-silicon-valley-influence-battle-for-ai-future/?sh=2c4b954e2dc4
Tomi Engdahl says:
Mike Wheatley / SiliconANGLE:
Cisco introduces new AI products and services, including its Nexus HyperFabric AI clusters in partnership with Nvidia, and announces a $1B AI investment fund — Cisco Systems Inc. is doubling down on its artificial intelligence push with a host of new AI-powered products and services, plus various other initiatives focused on AI.
Cisco advances intelligent networking and security as it announces $1B AI investment fund
https://siliconangle.com/2024/06/04/cisco-advances-intelligent-networking-security-announces-1b-ai-investment-fund/
Cisco Systems Inc. is doubling down on its artificial intelligence push with a host of new AI-powered products and services, plus various other initiatives focused on AI.
At its annual networking and security conference, Cisco Live 2024, the company today unveiled a new breed of AI-infused networking, observability, productivity and security offerings that stretch across its entire portfolio of products. It also announced the creation of a new, $1 billion global investment fund for AI startups that aims to build and expand on the most innovative AI systems around, plus new training initiatives to help partners and customers make the most of their AI investments.
The company kicked things off with its new AI networking products, saying they’re designed to make life easier for customers looking to implement the next generation of AI-powered applications. At the same time, the new offerings leverage AI themselves to help give customers greater visibility into their networks.
Tomi Engdahl says:
Gerrit De Vynck / Washington Post:
Pika, which offers AI tools to generate and edit videos, raised $80M from Spark, Lightspeed, and others at a $470M valuation, taking its total funding to $135M — Pika, a video-generation start-up that raised $80 million in its latest funding round, already has a publicly available product, as OpenAI and Google keep theirs under wraps
AI video start-ups race ahead as Big Tech competition looms
https://www.washingtonpost.com/technology/2024/06/04/pika-funding-openai-sora-google-video/
Pika, a video-generation start-up that raised $80 million in its latest funding round, already has a publicly available product, as OpenAI and Google keep theirs under wraps.
Tomi Engdahl says:
Ivan Mehta / TechCrunch:
OpenAI fixed two separate outages that impacted ChatGPT for multiple hours, starting at around 12:20am PDT; ChatGPT went down for over five hours on May 23
OpenAI fixes the issue that caused ChatGPT outage for several hours
https://techcrunch.com/2024/06/04/chatgpt-is-down-for-several-users-openai-is-working-on-a-fix/
TechCrunch Logo
Apps
OpenAI fixes the issue that caused ChatGPT outage for several hours
Ivan Mehta
9:00 AM PDT • June 4, 2024
Comment
OpenAI and ChatGPT logos
Image Credits: Didem Mente/Anadolu Agency / Getty Images
Popular AI chatbot service ChatGPT faced multiple outages today. OpenAI had to work for hours on these issues to make the service available to users again.
According to the company’s service status tracker, the company started working on the first issue at 12:21 AM PDT on June 4. OpenAI said it fixed this issue at 04.19 AM PT.
However, hours after this update, the company said it was investigating another outage at 07.33 AM PT. OpenAI posted an update at 10.17 AM PDT saying that all systems were operational.
“We experienced a major outage impacting all users on all plans of ChatGPT. The impact included all ChatGPT-related services. The impact did not include platform.openai.com or the API. This incident started June 4th at 2:15 PM GMT and was resolved June 4th at 5:01 PM GMT,” Open.AI said without providing any additional context for the outage.
Tomi Engdahl says:
Emilia David / The Verge:
Raspberry Pi partners with chipmaker Hailo for its AI Kit, an add-on for the Raspberry Pi 5 that runs Hailo’s Hailo-8L M.2 accelerator, available “soon” for $70
Even the Raspberry Pi is getting in on AI
/ Raspberry Pi partnered with Hailo to provide an optional AI add-on to its microcomputers.
https://www.theverge.com/2024/6/4/24170818/raspberry-pi-ai-chip-hailo-devices
Tomi Engdahl says:
Huoli pois, koodari koodaa edelleen
https://etn.fi/index.php/13-news/16292-huoli-pois-koodari-koodaa-edelleen
Tekoälyn yleistyminen erityisesti ohjelmoinnissa on saanut monet epäilemään, tarvikaanko ohjelmistokehittäjää tulevaisuudessa lainkaan. AWS Suomen ja Baltian maajohtaja Sari Uusitalon mukaan huoli on turha. – Koodari koodaa edelleen.
Amazonin pilvipalveluryhmä eli Amazon Web Services esitteli uusia tuotteitaan tällä viikolla Tukholmassa AWS Summit -tapahtumassa. Stockholmsmässanin parin vuoden päästä purettaviin tiloihin oli tullut yli 8600 ammattilaista. Eikä ollut epäselvää, mikä aiheutti messukeskuksessa eniten pöhinää. Se oli ilman muuta tekoäly.
Tekoäly on yhä keskeisemmässä roolissa AWS:n tuoteportfoliossa. Suomessa AWS:n asiakkaita ovat esimerkiksi Finnair, VR, sähköautojen latausdataa pilveen syöttävä Kempower, Amer ja Suunto, jonka liikkumisen seuranta tallentuu AWS:n palvelimille. Kone käyttää jo AWS Bedrockia GenAI-sovellusten kehittämiseen. Sari Uusitalon mukaan AWS ei millään muotoa ole vain suomalaisyritysten kumppani.
Vielä muutama vuosi sitten datanhallinnassa vannottiin hybridipilven nimeen. Oli tavoiteltavaa säilyttää osa datasta pilvessä ja osa yrityksen omilla palvelimilla. Tekoälyn myötä tämä malli on muuttumassa.
- Uskomme, että ajan myötä kaikki data siirtyy pilveen. Kehityksen vauhti toki riippuu toimialasta ja yrityksen tarpeista. AI on ollut boosteri tähän kehitykseen. Datan täytyy olla eheää. Myös strukturoimaton (unstructured) data on pilvessä, ja GenAI-mallien avulla sitäkin voidaan hyödyntää, Uusitalo kuvaa uutta normaalia.
Jo vuosien ajan on puhuttu siitä, että ”data on uusi öljy”. Tällä on viitattu yleensä Teollisuus 4.0:een ja IoT-datan hyödyntämiseen. Tekoälyn myötä datan merkitys vain korostuu. Joidenkin analyytikoiden mukaan AI-mallien virittäminen (finetuning) nousee yhdeksi softakehittäjän keskeisistä taidoista. Onnistuuko tämä muutos suomalaisilta?
Sari Uusitalo uskoo tähän vakaasti. Toki uudet taidot vaativat opiskelua ja uuden oppimista.
Koodareille AWS on tuomassa esimerkiksi Bedrock Studiota, joka on nyt preview-vaiheessa. – Studion tarkoitus on tehdä tekoälyn käyttöönotto mahdollisimman helpoksi. Siinä voi analysoida kielimalleja ja arvioida, mikä sopii parhaiten omiin tarkoituksiin. Kun kehittäjä saa käsiinsä apuvälineitä, koodaus tehostuu. Työkaluilla voidaan automatisoida rutiineja.
AWS:n datakeskukset jaetaan 33 Regioniin, joista 8 on Euroopassa. Suomalaisille yrityksille lähin regioona AWS-datalle on Tukholmassa
Aina dataa ei ehdi hakea Tukholmasta, mitä varten AWS:llä on Helsingillä paikalliskeskus. Tai local zone, kuten virtuaalista pilveä AWS:n kielessä nimitetään.
- Helsinki-Espoo -alueella se on, suostuu Uusitalo tarkentamaan.
Paikallisesti tallennettua dataa tarvitaan, jos sovellukselta vaaditaan alle 10 millisekunnin latenssia.
- Supercell on hyvä esimerkki. Pelaajan pelikokemus pitää saada mahdollisimman hyväksi. IoT-data on toinen. Ja pörssi, Uusitalo luettelee.
Tomi Engdahl says:
Future AI trends: the move from generative to interactive AI
Generative AI has helped to transform the customer experience. Interactive AI will revolutionise it.
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=IwAR2DN1C_3XgphRidT6oVXKptXr9GSM8LYJCkjj0Bim-65MRzSyYlxp6ZraA_aem_AaqjIsXpj4ZwSy3aZPKc8flAfSNgz6TvKoWA0jvMorCQCYB7efI_SbH_tS60NGIHUSo0DuFmPEwRUhmkSQu17uHK&utm_id=120206853243350278&utm_term=120209072778990278&utm_campaign=120206853243350278
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.
Where are we now?
Customers increasingly want two-way conversational experiences with a business or brand on the channels they use with their families and friends. They demand one seamless experience rather than a disparate journey across multiple devices and channels.
Businesses globally have responded to this demand and are implementing conversational experiences across the customer journey, whether for marketing, sales, or support.
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.
For instance, businesses have introduced generative AI chatbots trained on their company’s knowledge base, documentation, and call transcripts to provide risk-free conversational support. Other firms are deploying such chatbots on social media to monitor feedback, reply to comments and flag when human intervention is required.
As a result, brands that can provide consistent support to customers as they progress through purchase will see higher loyalty and customer satisfaction.
Tomi Engdahl says:
How is AI developing?
The future lies in interactive AI, where bots or intelligent agents can complete tasks set for them by interacting with other software or people to achieve those tasks.
Within the world of commerce and retail, the journey has just begun. Interactive AI will enable a federation of different bots and AI algorithms to trigger actions across the customer journey. More brands will integrate such processes across the customer journey to support marketing, sales and support automation, driving efficiencies while improving the customer experience.
For instance, a brand might boost sales by using an AI Assistant to fulfil routine tasks such as creating a Black Friday marketing campaign. A federation of bots might use generative AI to draft copy and create visuals, use another algorithm to segment customers, and a further AI chatbot to engage customers on their favourite chat app and answer customer questions. Finally, this kind of bot orchestration may use a further AI algorithm to optimise notifications and ensure customers are engaged to make purchases and drive sales in a secure and trustworthy way by eliminating hallucinations.
In short, the potential for interactive AI to further enhance and personalise the customer experience is vast and will usher in an ever more dynamic customer landscape.
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=IwAR2DN1C_3XgphRidT6oVXKptXr9GSM8LYJCkjj0Bim-65MRzSyYlxp6ZraA_aem_AaqjIsXpj4ZwSy3aZPKc8flAfSNgz6TvKoWA0jvMorCQCYB7efI_SbH_tS60NGIHUSo0DuFmPEwRUhmkSQu17uHK&utm_id=120206853243350278&utm_term=120209072778990278&utm_campaign=120206853243350278
Tomi Engdahl says:
Uncensor any LLM with abliteration
https://huggingface.co/blog/mlabonne/abliteration
Tomi Engdahl says:
DO WE DARE USE GENERATIVE AI FOR MENTAL HEALTH?
Woebot, a mental-health chatbot, is testing it out
https://spectrum.ieee.org/woebot
Tomi Engdahl says:
Organic Intelligence: A new metric, eliminating uncertainty in data analysis
https://researchfeatures.com/organic-intelligence-new-metric-eliminating-uncertainty-data-analysis/
All conventional data analysis methods involve uncertainty. These methods support Artificial Intelligence strategies, so the resulting uncertainty can be risky and dangerous.
OICT analyses raw data without assumptions or approximations. It offers better clarity by providing absolute certainty, eliminating the need for inaccurate learning and ‘best guesses’.
Importantly, this groundbreaking technological advancement assesses a true picture of reality.
Tomi Engdahl says:
https://tinybots.net/artbot/create
Tomi Engdahl says:
https://www.techradar.com/news/live/chatgpt-is-down-heres-what-we-know-about-the-outage-so-far
Tomi Engdahl says:
AutoCoder: The First Large Language Model to Surpass GPT-4 Turbo (April 2024) and GPT-4o in pass@1 on the Human Eval Benchmark Test (90.9% vs. 90.2%)
https://www.marktechpost.com/2024/05/31/autocoder-the-first-large-language-model-to-surpass-gpt-4-turbo-april-2024-and-gpt-4o-in-pass1-on-the-human-eval-benchmark-test-90-9-vs-90-2/
Tomi Engdahl says:
LLM360 Introduces K2: A Fully-Reproducible Open-Sourced Large Language Model Efficiently Surpassing Llama 2 70B with 35% Less Computational Power
https://www.marktechpost.com/2024/06/01/llm360-introduces-k2-a-fully-reproducible-open-sourced-large-language-model-efficiently-surpassing-llama-2-70b-with-35-less-computational-power/
Tomi Engdahl says:
Do you really need that GPU or NPU for your AI apps?
And how do we feel about Palantir selling its ML-based targeting system to the entire US military? Tune in
https://www.theregister.com/2024/06/01/kettle_ai/
Tomi Engdahl says:
Tekoälyhakuihin hukkautuu paljon energiaa
https://etn.fi/index.php/13-news/16301-tekoaelyhakuihin-hukkautuu-paljon-energiaa
Tekoälyn käytön lisääntyessä datakeskukset tulevat vaatimaan yhä enemmän virtaa vastatakseen kasvaviin prosessointivaatimuksiin. Onsemi on esitellyt uusia tehopiirejä, joiden avulla näiden halujen aiheuttamia tehohäviöitä voidaan pienentää.
Perinteiseen verkkohakuun verrattuna tekoälyä tukevan hakukoneen haku vaatii yli kymmenkertaisesti enemmän tehoa, minkä takia datakeskusten tehontarpeen odotetaan nousevan maailmanlaajuisesti arviolta 1000 terawattituntiin alle kahdessa vuodessa. Yhden tekoälyn tukeman pyynnön käsittelemiseksi energiaa muunnetaan neljä kertaa verkosta prosessoriin, mikä voi johtaa noin 12 prosentin energiahäviöön.
Tomi Engdahl says:
Amanda Silberling / TechCrunch:
Cara, an artist-run, anti-AI social platform, grew from 40K to 650K users in the past week, as some Instagram artists leave over plans to train AI on user data — ‘Building a product is a bit like making art,’ said founder Jingna Zhang — Artists have finally had enough with Meta’s predatory AI policies …
A social app for creatives, Cara grew from 40k to 650k users in a week because artists are fed up with Meta’s AI policies
‘Building a product is a bit like making art,’ said founder Jingna Zhang
https://techcrunch.com/2024/06/06/a-social-app-for-creatives-cara-grew-from-40k-to-650k-users-in-a-week-because-artists-are-fed-up-with-metas-ai-policies/
Tomi Engdahl says:
Zac Bowden / Windows Central:
After Microsoft eroded Windows users’ trust with bad practices for years, Recall is a PR disaster, as users remain skeptical despite the company’s assurances — The world is up-in-arms over Windows Recall, but why? It stems from Microsoft’s seeming lack of care for Windows and its users.
A PR disaster: Microsoft has lost trust with its users, and Windows Recall is the straw that broke the camel’s back
News
By Zac Bowden
https://www.windowscentral.com/software-apps/windows-11/microsoft-has-lost-trust-with-its-users-windows-recall-is-the-last-straw
The world is up-in-arms over Windows Recall, but why? It stems from Microsoft’s seeming lack of care for Windows and its users.
Tomi Engdahl says:
Aisha Counts / Bloomberg:
Meta plans WhatsApp features, including an AI-powered assistant for businesses in India and Singapore, and AI tools to create ads in English-speaking markets — – Assistant lets businesses chat with customers using AI — New features will roll out first in India and Singapore
Meta’s WhatsApp Is Adding a New AI Assistant for Businesses
https://www.bloomberg.com/news/articles/2024-06-06/meta-s-whatsapp-is-adding-a-new-ai-assistant-for-businesses
Assistant lets businesses chat with customers using AI
New features will roll out first in India and Singapore
Tomi Engdahl says:
New York Times:
Sources: as the Ai Pin flopped, with just ~10K orders by early April, Humane started talking to HP about selling itself for $1B+; Humane has raised $240M
https://www.nytimes.com/2024/06/06/technology/humane-ai-pin.html
Tomi Engdahl says:
Carl Franzen / VentureBeat:
Adobe faces a backlash over its terms of service that say Adobe may access user content via “automated and manual methods” to improve its services and software — Adobe has been one of the leading legacy enterprise software companies to embrace generative AI and make it accessible …
Adobe responds to vocal uproar over new Terms of Service language
https://venturebeat.com/ai/adobe-responds-to-vocal-uproar-over-new-terms-of-service-language/
Tomi Engdahl says:
SAP työntää copilottinsa kaikkiin tuotteisiin
https://etn.fi/index.php/13-news/16306-sap-tyoentaeae-copilottinsa-kaikkiin-tuotteisiin
Saksalaisen SAP:n toiminnanohjausjärjestelmää on moitittu raskaaksi ja jopa hankalaksi, mutta ehkäpä siitäkin tulee ketterämpi tekoälyn ansiosta. Jokavuotisessa SAP Sapphire -kehittäjäkokouksessaan yritys kertoi, että sen tekoälyä hyödyntävä copilot-avustin tuodaan osaksi kaikkia työkaluja.
SAP:n copilot on nimeltään Joule. Pääjohtaja Christian Kleinin (kuvassa) mukaan perustason tekoälyratkaisut sisältyvät jo nykyisiin lisenssimaksuihin. Jos asiakas käyttää arvokkaampia ratkaisuja kuten GenAI-tekniikkaa, siitä tullaan laskuttamaan enemmän.
Tomi Engdahl says:
Tekoäly mullistaa sen, miten sovelluksia rakennetaan
https://etn.fi/index.php/13-news/16305-tekoaely-mullistaa-sen-miten-sovelluksia-rakennetaan
Tekoälyä voidaan pitää neljäntenä isona vallankumouksena. Sitä ei kuitenkaan pidä ajatella työkaluna, joka osaa kirjoittaa runoja. – Tekoäly muuttaa fundamentaalisesti sen, miten sovelluksia ja palveluita rakennetaan, sanoi Amazon Web Servicesin Becky Weiss yhtiönsä AWS Summitissa tällä viikolla.
Huonosti kääntyvällä Vice President Principal Engineer -tittelillä toimivan Weissin mukaan tekoäly tarkoittaa datan tulkintaa uudella tavalla. Ennen kaikkea se tarkoittaa kykyä automatisoida kompleksisia tehtäviä. – AWS:n tehtävä on tuoda tämä kaikki laajasti ja helposti saataville, Weiss sanoi keynotessaan.
Tällä hetkellä käynnissä on jonkinlaisen suurten LLM-kielimallien välinen taistelu. Eri yritykset kisaavat siitä, kenen malli on nopein ja tehokkain. Weiss muistuttaa, ettei yksikään malli ainoana riitä kaikkeen. Tämän takia aAWS kehittää työkalua, jonka avulla pilviasiakas voi valita omaan sovellukseen sopivimman mallin.
Reilu vuosi sitten AWS lanseerasi asiakkailleen Bedrock-palvelun, jossa nämä voivat rakentaa, ottaa käyttöön ja skaalata koneoppimismalleja omaan käyttöönsä. Tarjolla on laaja valikoima malleja, joista valita. Becky Weissin mukaan keskeisintä on kuitenkin edelleen data.
Tomi Engdahl says:
Meet Tsinghua University’s GLM-4-9B-Chat-1M: An Outstanding Language Model Challenging GPT 4V, Gemini Pro (on vision), Mistral and Llama 3 8B
https://www.marktechpost.com/2024/06/05/meet-tsinghua-universitys-glm-4-9b-chat-1m-an-outstanding-language-model-challenging-gpt-4v-gemini-pro-on-vision-mistral-and-llama-3-8b/
Tomi Engdahl says:
https://etn.fi/index.php/13-news/16306-sap-tyoentaeae-copilottinsa-kaikkiin-tuotteisiin
Tomi Engdahl says:
Artificial Intelligence
Microsoft Bows to Public Pressure, Disables Controversial Windows Recall by Default
Amidst public pressure, Microsoft changes the set-up experience of Copilot+ PCs to disable the controversial Windows Recall feature by default.
https://www.securityweek.com/microsoft-bows-to-public-pressure-disables-controversial-windows-recall-by-default/
Tomi Engdahl says:
EMO: Alibaba’s Diffusion Model-Based Talking Portrait Generator
https://hackaday.com/2024/06/10/emo-alibabas-diffusion-model-based-talking-portrait-generator/
Alibaba’s EMO (or Emote Portrait Alive) framework is a recent entry in a series of attempts to generate a talking head using existing audio (spoken word or vocal audio) and a reference portrait image as inputs. At its core it uses a diffusion model that is trained on 250 hours of video footage and over 150 million images. But unlike previous attempts, it adds what the researchers call a speed controller and a face region controller. These serve to stabilize the generated frames, along with an additional module to stop the diffusion model from outputting frames that feature a result too distinct from the reference image used as input.
In the related paper by [Linrui Tian] and colleagues a number of comparisons are shown between EMO and other frameworks, claiming significant improvements over these.
EMO: Emote Portrait Alive – Generating Expressive Portrait Videos with Audio2Video Diffusion Model under Weak Conditions
https://humanaigc.github.io/emote-portrait-alive/
Tomi Engdahl says:
AI Kayak Controller Lets The Paddle Show The Way
https://hackaday.com/2024/06/10/ai-kayak-controller-lets-the-paddle-show-the-way/
Tomi Engdahl says:
Despite modest intellectual improvements, AGI’s human-like behaviour raises serious ethical concerns. Shannon Vallor argues that AI today only presents the illusion of consciousness.
https://iai.tv/articles/the-dangerous-illusion-of-ai-consciousness-auid-2847?utm_source=Facebook%20&utm_medium=Paid&utm_campaign=Science&utm_content=AI%20Consciousness
Tomi Engdahl says:
Usama Jawad / XDA Developers:
Microsoft is removing the Custom GPT Builder feature from the consumer version of Copilot Pro on July 10, just three months after launching the feature — Microsoft has decided to retire GPT Builder in Copilot Pro for consumers, just three months after it began rolling out the service broadly.
https://www.xda-developers.com/microsoft-killing-gpt-builder-copilot-pro-consumers-availability/