Täällä seurataan AI-ripuligeneraattoreiden maailmanvalloitusta

[karhuherra baardington]

Tulee robocop mikä suoltaa rasistista paskaa eli ei nyt niin paljoa eroa ihmispoliisista

[pigra senlaborulo]

demokraatti.fi
Tekoälyn sotilaallisesta käytöstä poliittinen julistus – Suomi liittyi mukaan
DEMOKRAATTI/STT Demokraatti
~2 minutes

Suomi on liittynyt tekoälyn vastuullista sotilaallista käyttöä koskevaan poliittiseen julistukseen. Yhdysvaltojen johdolla valmisteltu julistus julkaistiin maanantaina New Yorkissa.

Puolustusministeri Antti Häkkäsen (kok.) mukaan tekoälyn nopea kehittyminen sotilasteknologiassa on luonut tarpeen sääntelyn aikaansaamiselle.

– Tekoäly ja sen mahdollistama koneautonomia ovat tulleet yhä tärkeämmiksi paitsi varsinaisissa asejärjestelmissä, myös esimerkiksi tilannekuvan tuottamisessa, suunnittelussa ja logistiikassa.

Häkkäsen mukaan tekoälyn sotilaallisessa käytössä olennaista on sen hyödyntäminen moraalisesti, eettisesti ja juridisesti oikein. Tekoälyyn nojautuvat järjestelmät on suunniteltava ja testattava oikein suhteessa tehtäväänsä, minkä lisäksi niiden käyttö on vastuutettava ja rajoitettava selkeän komentoketjun mukaisesti.

-  Osallistuvat valtiot sitoutuvat kymmeneen periaatteeseen, joilla edistetään vastuullisuutta ja turvallisuutta tekoälyn ja autonomisten toimintojen sotilaallisessa kehittämisessä ja käytössä. Ne korostavat kansainvälisen oikeuden ja kansainvälisen humanitaarisen oikeuden periaatteiden noudattamista, hän toteaa puolustusministeriön tiedotteessa.

KYSEESSÄ on poliittinen sitoumus, joka ei luo julistukseen liittyneille uusia oikeudellisia velvoitteita.

Häkkäsen mukaan tekoälyn sotilaalliseen käyttöön liittyy eettisten kysymysten lisäksi myös puolustusteollisia ja kauppapoliittisia mahdollisuuksia, sillä tekoälyn ja koneautonomian osaamispohja on Suomessa vahva.

-  Tekoälyn ja autonomisten asejärjestelmien sääntelyn avulla kehitämme omaa puolustuskykyämme yhdessä liittolaistemme kanssa.

https://demokraatti.fi/tekoalyn-sotilaa ... yi-mukaan/

[pigra senlaborulo]

mikrobitti.fi
Meta ilmoitti rajoituksista tekoälytyökaluihin – näitä aiheita ei sallita
Petri Ranta
~2 minutes

Facebookin ja Instagramin omistaja Meta on ilmoittanut, ettei sen tekoälytyökaluja saa käyttää poliittisen kampanjamateriaalin luomiseen.

Tavoitteena on estää työkalujen käyttäjiä mahdollisesti luomasta niillä vaaleihin ja niiden ehdokkaisiin liittyvää valetietoa.

Metan tekoälytyökalut tuotiin käyttöön rajoitetulle joukolla mainostajia vuoden 2023 alussa.

Työkaluilla voi muun muassa luoda taustakuvia ja luoda kuviin yksityiskohtia yksinkertaisen tekstikuvauksen avulla.

Lokakuun alussa Meta ilmoitti, että työkalut on tarkoitus julkaista kaikille mainostajille maailmanlaajuisesti ensi vuoden alussa, Reuters kertoo.

Poliittiset kampanjat eivät ole ainoa asia, joiden käyttöön Meta ei ole työkaluaan antamassa. Niitä ei saa käyttää myöskään esimerkiksi terveydenhoitoon tai taloudellisiin palveluihin liittyvän materiaalin luomisessa.

Lausunnossaan Meta ilmoittaa, että sen lähestymistapa generatiivista tekoälyä kohtaan toivottavasti auttaa uhkakuvien paremmassa ymmärtämisessä ja rajoitusten luomisessa.

Meta on tiettävästi luonut ohjeen myös siihen, miten sen alustoilla jatkossa suhtaudutaan muilla tekoälytyökaluilla luotuun sisältöön. The Verge kertoo, että yhtiö aikoo jatkossa vaatia realistisen tekoälysisällön yhteydessä maininnan siitä, ettei sisältö ole todellista.

Metan tavoin useat muut teknologia-alan jättiyhtiöt ovat alkaneet kehittää omia tekoälytyökalujaan OpenAI:n viime syksynä julkaistun ChatGPT:n rohkaisemana. Useimmat muut yritykset eivät kuitenkaan ole toistaiseksi tehneet linjanvetoa siitä, millaisia rajoituksia työkalujen käytössä on.

Esimerkiksi Googlen omistava Alphabet ilmoitti marraskuun alussa omista tekoälytyökaluistaan. Yhtiön mukaan tiettyjä ”poliittisia” sanoja ei kuitenkaan hyväksytä tekstisyötteessä.

https://www.mikrobitti.fi/uutiset/mb/00 ... 6f076c602a

[Spandau Mullet]

https://www.404media.co/andreessen-horo ... fake-porn/

Andreessen Horowitz Invests in Civitai, Which Profits From Nonconsensual AI Porn

Andreessen Horowitz, also known as a16z, the influential Silicon Valley venture capital firm that was an early investor in Facebook, Lyft, and other tech giants, has invested in Civitai, a giant platform for sharing AI models that enables and profits from the creation of AI generated nonconsensual sexual images of real people. That includes launching a feature where people can list “bounties” for others to create AI models of specific targets.

Civitai said that it raised $5.1 million in a seed funding round led by a16z.

A16z’s official website, which includes a jobs board with open positions at companies in its portfolio, currently lists five jobs at Civitai. According to a16z’s site, these jobs were posted more than 30 days ago.

A16z regularly announces investments the company is making on its site, but has not publicly announced its investment in Civitai yet. Civitai also gives its community regular updates on the state of the platform and company (it recently celebrated its one year anniversary, for example) in blogs and on its official Discord, but has not announced taking any investment.

A16z did not respond to a request for comment. When asked about a16z’s investment in Civitai over Discord, a community engagement manager at Civitai told 404 Media that “There will be a press release/announcement shortly.” Civitai then published a press release confirming the investment minutes after 404 Media reached out for comment.

In August, 404 Media published an investigation into Civitai, which explained how the platform works, and enables the creation of AI-generated nonconsensual sexual images, and profits from it. Civitai allows users to share modified models of the open source text-to-image AI tool Stable Diffusion. These modified models are often trained on images of celebrities, influencers, YouTubers, and athletes, almost exclusively women, to recreate their likeness. Those models can then be combined with AI models that are trained on porn in order to instantly generate nonconsensual sexual images.

Technically, Civitai’s policy forbids users from generating nonconsensual porn. Models that generate pornographic images are allowed. Models that generate images of real people are allowed. Combining the two is not. But there’s nothing preventing people from putting the pieces together, generating non-consensual sexual images, and sharing them off CivitAI’s platform.

While 404 Media has seen CivitAI enforce this policy and remove such content multiple times, non-consensual sexual imagery is still posted to the site regularly, and in some cases has stayed online for months.

Since 404 Media published its investigation in August, Civitai’s nonconsensual AI porn problem has only gotten worse. Previously, Civitai sent users to other generative AI platforms in order to actually use the models to create images. Today, Civitai has its own web interface for using the models to generate images directly on its site. 404 Media was able to easily and instantly use this feature to generate nonconsensual sexual images of a real person by searching the name of a celebrity, selecting that AI model, selecting an AI model designed to generate porn, and writing a pornographic prompt.

As 404 Media reported earlier this week, Civitai has also recently introduced a “bounties” feature. These bounties allow users to ask the Civitai community to create AI models that generate images of specific styles, compositions, or specific real people, and reward the best AI model that does so with a virtual currency called “Buzz” that users can buy with real money.

404 Media has found that there are dozens of bounties for celebrities, Instagram influencers, and Twitch streamers. But 404 Media has seen at least one bounty for a private person who has no significant public online presence.

“Civitai is the prime example of a company that’s already built an incredible, engaged community and all without spending a dime on marketing,” Brian Kim, a partner at a16z, said in the press release. “Our investment in the company will only supercharge something that’s already working incredibly well, by contributing to a world where any individual can take advantage of what the biggest technological shift of our generation – AI – has to offer.”

A16z has mentioned Civitai several times before 404 Media learned about its investment in the company. In September, it published a blog analyzing the generative AI market, which found that Civitai was the seventh most popular tool in that sector. The blog did not mention adult content on Civitai or that nonconsensual sexual images were a major driving force for the platform. In June, a16z published a blog that noted users could use Civitai to create “DIY” companions. “[AI models] on Civitai give you granular control over the style, detail, and setting of images of your AI partner,” the blog said.

That a16z would invest in an AI company that is actively causing harm to real people is not surprising given its stated approach to this technology. In October, Marc Andreessen, half the firm’s namesake, published a “manifesto” in which he names the “enemies” of technological progress, and argues that any attempt to regulate AI and its potential harms “is a form of murder.”

Vittu tää on kyllä paska cyberpunk-tulevaisuus.

[pigra senlaborulo]

YouTube Now Requires Creators To Label AI Content

As AI-generated content continues to encroach on everything from advertising to the voice acting profession, YouTube is adding a requirement for users to flag their videos when they include anything made by an AI program. However, looking at the guidelines, it doesn’t seem like the video hosting site has any way to actually enforce or detect this.

YouTube’s Product Management Vice Presidents, Jennifer Flannery O’Connor and Emily Moxley, broke down the new policy in a blog post on November 14. First and foremost, any video that contains AI-generated content will require disclosure and content labels in the video description that make it clear aspects of the video were created by AI. The examples given include a video that “realistically depicts an event that never happened” as well as deepfakes showing an individual “saying or doing something they didn’t actually do.”

The blog post says this new policy is meant to help combat misinformation, especially regarding real-world issues like elections and ongoing health and world crises. It also states that some AI-generated content, whether it’s labeled or not, may be removed from YouTube if the disclaimer “may not be enough to mitigate the risk of harm.” An example of this given by YouTube is a realistic portrayal of violence that solely exists to gross people out, as opposed to a historical video of the educational or informative sort that also includes violence.

Alongside the disclaimer, YouTube is rolling out community guidelines that will allow those affected by AI-generated content to request videos be removed on those grounds. So if someone is using AI to simulate you doing something you didn’t do, you can request to have those videos removed, with YouTube offering the specific example of musicians whose voices are being mimicked by AI software.

One distinction made is that if AI-generated voices are part of an analysis, such as a creator discussing the trend of AI covers and including audio that sounds like a singer performing someone else’s song, the video may not be taken down. But it sounds like videos that are just songs performed by an AI imitating someone’s voice can be taken down at an artist’s request. Parody or satire is also, apparently, fair game.

The big question here is whether or not YouTube actually has any means of enforcing this beyond the threat of consequences, including “content removal, suspension from the YouTube Partner Program, or other penalties” for those who consistently fail to disclose. Presumably the “other penalties” could mean an eventual ban from the platform, but even so, it sounds as if the entire thing is currently self-imposed and working on an honor system.

While there might be some kinks to work out here, it is a relief to see some work being done on huge platforms to combat the misinformation brought on by AI tools. I spend a lot of time on TikTok and while AI covers and other audio have become prominent on that platform, I’ve anecdotally seen a lot of users make entire accounts that do nothing but churn out AI content without disclosing it at all. I’m a chronic scroller, so I’ve learned the signs to look and listen for, but as AI tools become more and more widespread, it’s becoming more and more likely that people who don’t know better will start to take these videos at face value.

https://kotaku.com/youtube-ai-disclosure-1851020956

[pigra senlaborulo]

mikrobitti.fi
Oletko jo koukussa tekoälyn käyttämiseen? Microsoft saattaa tehdä siitä lopun
Joakim Kullas
~2 minutes

Microsoft on sisällyttänyt verkkopalveluidensa käyttöehtoihin varoituksen siitä, että liiallinen generatiivisen tekoälyn käyttö voi johtaa pääsyn rajoittamiseen tekoälypalveluihin. Uusi varoitus on kirjattu käyttöehtoihin ”kapasiteettirajoitukset”-otsikon alle, uutisoi The Register.

Käyttöehdoissa ei kuitenkaan tarkenneta, mitä tekoälyn liiallinen käyttö tarkoittaa, kuinka pitkään käyttäjien pääsyä tekoälypalveluihin voidaan rajoittaa tai miten pääsyn rajoittaminen käytännössä toteutetaan.

Microsoftin tarve kaventaa tekoälyn käyttöä voi kertoa siitä, että yhtiön tekoälyarkkitehtuurissa on pullonkauloja, jotka voivat vähentää palvelun tehoa osalla käyttäjistä. Ongelma olisi yhtiölle merkittävä, sillä ajoittain huonosti toimivat tekoälypalvelut saattavat ajaa asiakkaita kilpailijoiden ohjelmistojen pariin.

Toinen selitys tekoälyn käytön rajoittamiselle voi olla se, että Microsoft haluaa karsia sen pyörittämiseen tarvittavia kustannuksia. Yhtiö on jo varoittanut sijoittajia siitä, että tekoälyn skaalaaminen tulee kalliiksi.

Microsoft kertoi 24. lokakuuta järjestämässään tulospuhelussa, että vuoden 2024 ensimmäisen kvartaalin kustannukset kohosivat 11,2 miljardiin dollariin, josta osa kuluu tekoälyinfrastruktuurin skaalaamiseen. Yhtiö huomautti, että tekoälyn kasvavaan kysyntään vastaaminen skaalaamalla näkyy myös sen katteessa.

Korkeat kustannukset johtuvat siitä, että tekoälyn voimanpesinä toimivat palvelimet ja grafiikkasuorittimet ovat kalliita hankkia ja operoida. Näin ollen tekoälyn liiallisen käytön rajoittaminen voi tuoda Microsoftille merkittäviä säästöjä.

https://www.mikrobitti.fi/uutiset/mb/cd ... ccae5cbb92

[hellästi takaluukkuun pantu koira]

jotain hyvääkin, googlen koneoppimismalli tekee kymmenen päivän sääennusteita nopeammin ja yleensä tarkemmin kuin euroopan keskipitkien sääennusteiden keskuksen numeeriset mallit. ennusteen erotuskyky on kuitenkin huonompi.

https://arstechnica.com/?p=1983760

[Spandau Mullet]

jotain hyvääkin, googlen koneoppimismalli tekee kymmenen päivän sääennusteita nopeammin ja yleensä tarkemmin kuin euroopan keskipitkien sääennusteiden keskuksen numeeriset mallit. ennusteen erotuskyky on kuitenkin huonompi.

https://arstechnica.com/?p=1983760

Täähän vaikuttaakin ihan kunnon AI-sovellukselta eikä miltään bullshit-generaattorilta.

[pigra senlaborulo]

Laulaja Édith Piafin elämästä kertovassa uudessa elokuvassa käytetään tekoälyä, jonka avulla ranskalaislaulaja saa kertoa oman tarinansa omalla äänellään. Asiasta kertovat levy-yhtiö Warner Music ja Piafin perikunta uutistoimisto Reutersin mukaan.

Ranskan Warner Musicin mukaan elokuva Edith on viimeistelyvaiheessa. Julkaisupäivää ei ole vielä kerrottu.

Levy-yhtiön tiedotteen mukaan kyseessä on ”innovatiivinen ja vallankumouksellinen teknologinen projekti, jossa käytetään tekoälyä Piafin äänen ja kuvan luomiseen uudelleen”.

Édith Piaf on Ranskan tunnetuin laulaja ja chansonin suuri tähti, jonka tunnetuimpia kappaleita ovat La vie en rose, Non, je ne regrette rien, Milord ja L’hymne à l’amour.

Lokakuun 11. päivänä hänen kuolemastaan tuli kuluneeksi 60 vuotta.

Piafin tarina Pariisin kaduilta juhlituksi kansallissankariksi on kerrottu populaarikulttuurissa jo monesti niin teatterien lavoilla kuin valkokankaillakin.

Tunnetuin teos lienee vuonna 2007 julkaistu elokuva Pariisin varpunen, jossa Piafia näytteli ranskalainen Marion Cotillard. Hän sai roolistaan Oscar-palkinnon.

Uusin Piaf-elokuva on teknisessä mielessä jatkoa The Beatles -yhtyeen marraskuun alussa julkaistulle singlelle Now And Then. Kappale luotiin vanhan kasettinauhoituksen pohjalta tekoälyn avulla.

Tekoäly auttoi eristämään John Lennonin laulun, jonka hän äänitti vuonna 1978, kaksi vuotta ennen kuin hänet murhattiin. Yhtyeen toisen kuolleen jäsenen George Harrisonin osuudet taas ovat peräisin 1990-luvun sessionauhoituksista. Paul McCartney ja Ringo Starr lisäsivät omat osuutensa tämän materiaalin päälle.

Viime viikolla single nousi Britannian singlelistan ykköseksi. Kyseessä on ensimmäinen Beatlesin ykköspaikka listalla 54 vuoteen.

Nyt myös Ranskan chansonin ykkösnimi on herätetty eloon tekoälyn sekä vanhojen ääni- ja kuvatallenteiden avulla. Tekoäly sai käsiteltäväkseen satoja, osin yli 80 vuotta vanhoja kuvia ja äänitteitä.

Puolitoista tuntia kestävä elokuva kertoo Piafin elämästä 1920–1960-luvuilla niin Pariisissa kuin New Yorkissakin. Tarinaa kertoo Piafin oma, tekoälyllä luotu ääni.

Piafin suurimmat hitit kuulostavat elokuvassa siltä kuin ennenkin. Niissä on käytetty hänen alkuperäisiä levytyksiään.

https://www.hs.fi/kulttuuri/art-2000009992140.html

[pigra senlaborulo]

mikrobitti.fi
OpenAI veti hätäjarrusta: maksulliset ChatGPT-tilaukset jäihin
Samuli Leppälä
~1 minute

OpenAI:n toimitusjohtaja Sam Altman on kertonut X-tilillään (entinen Twitter) tekoäly-yhtiön toistaiseksi jäädyttäneen mahdollisuuden hankkia kuukausimaksullisen ChatGPT Plus -tilauksen.

Syy tähän on varsin yksinkertainen. Noin viikko sitten OpenAI esitteli omassa DevDay-tapahtumassaan GPT Storen, jonne kuukausitilaajat voisivat kehittää omia versioitaan ChatGPT-tekoälybotista.

Yhteisön luomilla tekoälyvarianteilla on mahdollista lyödä rahoiksi, sillä OpenAI lupasi maksaa tarkemmin määrittelemättömiä summia tahoille, jotka kehittävät suosittuja ChatGPT-versiointeja.

Altmanin mukaan DevDayn jälkeen nousseet palvelun käyttäjämäärät ovat yksinkertaisesti ylittäneet reservissä olleen kapasiteetin. Tämän myötä uusien tilauksien tekeminen on toistaiseksi asetettu jäihin.

Toistaiseksi ei ole tietoa siitä, miten pitkään uudet tilaukset ovat jäässä. Tilaus maksaa 20 dollaria kuukaudessa eli noin 18,5 euroa.
https://www.mikrobitti.fi/uutiset/mb/87 ... 04cf747b67

[pigra senlaborulo]

A former EverQuest producer, WoW vets and Elden Ring devs are making an MMO filled with AI-generated NPCs

With a load of blockchain to boot


A bunch of MMO veterans led by one of the original producers of genre granddaddy EverQuest are working on a new online game that will use AI tools to create characters and have them interact with players. There's also mention of blockchain-like asset ownership and ambitions of being the next metaverse.


Avalon’s team is headed up by Jeffrey Butler, who was a producer on the original EverQuest before serving as lead on its cancelled reboot EverQuest Next. Joining Butler at Avalon studio Avalon - yes, the company and game are called the same thing - is apparently a team of industry vets made up of more EverQuest alumni and devs with credits on World of Warcraft, Call of Duty, Elden Ring and Diablo.

Avalon: The Game’s headline feature is that it’ll span a variety of settings and genres, from fantasy to cyberpunk, and let players hop between them while contributing to the game’s own setting and lore as they add to the world.

Helping to create those many worlds is a load of AI-powered tech behind the scenes, including procedural generation character creation tool Popul8 - which churns out a bunch of variations on NPC character models, and was recently used to make all those overly-teethy civilians in Cities: Skylines 2.


Even deeper down the AI hole is Avalon’s use of Inworld AI, which acts as a sort of built-in ChatGPT for NPCs’ interaction with players, generating dialogue and personalities for characters based on prompts and details given to them by devs.

The result of all this tech, at least going by the game’s brief teaser trailer, appears to be a generic-looking MMO with unremarkable snippets of combat that switch between bad-looking hack-and-slash fantasy to bad-looking third-person shooting against sci-fi mechs. There’s also a bit of Sims-like house building and city-building-esque layout of a settlement shown, but mostly it looks like one of those video games that is only played by fictional characters in a TV show. Wow, AI really is the future.

If that wasn’t enough to sour the taste, there’s also mention of a virtual economy and “digital sovereignty” that will let players own assets and worlds and transfer them between worlds and other games and platforms, all “authenticated by Avalon through the DAO”, which very much smacks of blockchain integration.

To complete your bingo card of off-putting buzzwords, a comment from Avalon: The Company SEO Sean Pinnock adds that the MMO aims to “make something that fulfills the promise of the metaverse” by letting players build the world for themselves.

If the crashing-and-burning of pretty much everything to have aimed at becoming the next “metaverse” or touching the blockchain is anything to go by, Avalon’s got some work ahead of it to deliver on its lofty ambitions. Work that its creators would apparently prefer AI or players to do, it seems.
https://www.rockpapershotgun.com/a-form ... rated-npcs

[pigra senlaborulo]

www.theverge.com
Microsoft is finally making custom chips — and they’re all about AI
8 - 10 minutes

The rumors are true: Microsoft has built its own custom AI chip that can be used to train large language models and potentially avoid a costly reliance on Nvidia. Microsoft has also built its own Arm-based CPU for cloud workloads. Both custom silicon chips are designed to power its Azure data centers and ready the company and its enterprise customers for a future full of AI.

Microsoft’s Azure Maia AI chip and Arm-powered Azure Cobalt CPU are arriving in 2024, on the back of a surge in demand this year for Nvidia’s H100 GPUs that are widely used to train and operate generative image tools and large language models. There’s such high demand for these GPUs that some have even fetched more than $40,000 on eBay.

“Microsoft actually has a long history in silicon development,” explains Rani Borkar, head of Azure hardware systems and infrastructure at Microsoft, in an interview with The Verge. Microsoft collaborated on silicon for the Xbox more than 20 years ago and has even co-engineered chips for its Surface devices. “These efforts are built on that experience,” says Borkar. “In 2017, we began architecting the cloud hardware stack and we began on that journey putting us on track to build our new custom chips.”

The new Azure Maia AI chip and Azure Cobalt CPU are both built in-house at Microsoft, combined with a deep overhaul of its entire cloud server stack to optimize performance, power, and cost. “We are rethinking the cloud infrastructure for the era of AI, and literally optimizing every layer of that infrastructure,” says Borkar.

The first two custom silicon chips designed by Microsoft for its cloud infrastructure.
The first two custom silicon chips designed by Microsoft for its cloud infrastructure.

Image: Microsoft

The Azure Cobalt CPU, named after the blue pigment, is a 128-core chip that’s built on an Arm Neoverse CSS design and customized for Microsoft. It’s designed to power general cloud services on Azure. “We’ve put a lot of thought into not just getting it to be highly performant, but also making sure we’re mindful of power management,” explains Borkar. “We made some very intentional design choices, including the ability to control performance and power consumption per core and on every single virtual machine.”

Microsoft is currently testing its Cobalt CPU on workloads like Microsoft Teams and SQL server, with plans to make virtual machines available to customers next year for a variety of workloads. While Borkar wouldn’t be drawn into direct comparisons with Amazon’s Graviton 3 servers that are available on AWS, there should be some noticeable performance gains over the Arm-based servers Microsoft is currently using for Azure. “Our initial testing shows that our performance is up to 40 percent better than what’s currently in our data centers that use commercial Arm servers,” says Borkar. Microsoft isn’t sharing full system specifications or benchmarks yet.

Microsoft’s Maia 100 AI accelerator, named after a bright blue star, is designed for running cloud AI workloads, like large language model training and inference. It will be used to power some of the company’s largest AI workloads on Azure, including parts of the multibillion-dollar partnership with OpenAI where Microsoft powers all of OpenAI’s workloads. The software giant has been collaborating with OpenAI on the design and testing phases of Maia.

“We were excited when Microsoft first shared their designs for the Maia chip, and we’ve worked together to refine and test it with our models,” says Sam Altman, CEO of OpenAI. “Azure’s end-to-end AI architecture, now optimized down to the silicon with Maia, paves the way for training more capable models and making those models cheaper for our customers.”

Manufactured on a 5-nanometer TSMC process, Maia has 105 billion transistors — around 30 percent fewer than the 153 billion found on AMD’s own Nvidia competitor, the MI300X AI GPU. “Maia supports our first implementation of the sub 8-bit data types, MX data types, in order to co-design hardware and software,” says Borkar. “This helps us support faster model training and inference times.”

Microsoft is part of a group that includes AMD, Arm, Intel, Meta, Nvidia, and Qualcomm that are standardizing the next generation of data formats for AI models. Microsoft is building on the collaborative and open work of the Open Compute Project (OCP) to adapt entire systems to the needs of AI.

A probe station used to test Microsoft’s Azure Cobalt system-on-chip.
A probe station used to test Microsoft’s Azure Cobalt system-on-chip.

Image: Microsoft

“Maia is the first complete liquid cooled server processor built by Microsoft,” reveals Borkar. “The goal here was to enable higher density of servers at higher efficiencies. Because we’re reimagining the entire stack we purposely think through every layer, so these systems are actually going to fit in our current data center footprint.”

That’s key for Microsoft to spin these AI servers up more quickly without having to make room for them in data centers around the world. Microsoft built a unique rack to house Maia server boards in, complete with a “sidekick” liquid chiller that works like a radiator you’d find in your car or a fancy gaming PC to cool the surface of the Maia chips.

Along with sharing MX data types, Microsoft is also sharing its rack designs with its partners so they can use them on systems with other silicon inside. But the Maia chip designs won’t be shared more broadly, Microsoft is keeping those in-house.

Maia 100 is currently being tested on GPT 3.5 Turbo, the same model that powers ChatGPT, Bing AI workloads, and GitHub Copilot. Microsoft is in the early phases of deployment and much like Cobalt it isn’t willing to release exact Maia specifications or performance benchmarks just yet.

The Maia 100 server rack and “sidekick” cooling.
The Maia 100 server rack and “sidekick” cooling.

Image: Microsoft

That makes it difficult to decipher exactly how Maia will compare to Nvidia’s popular H100 GPU, the recently announced H200, or even AMD’s latest MI300X. Borkar didn’t want to discuss comparisons, instead reiterating that partnerships with Nvidia and AMD are still very key for the future of Azure’s AI cloud. “At the scale at which the cloud operates, it’s really important to optimize and integrate every layer of the stack, to maximize performance, to diversify the supply chain, and frankly to give our customers infrastructure choices,” says Borkar.

That diversification of supply chains is important to Microsoft, particularly when Nvidia is the key supplier of AI server chips right now and companies have been racing to buy up these chips. Estimates have suggested OpenAI needed more than 30,000 of Nvidia’s older A100 GPUs for the commercialization of ChatGPT, so Microsoft’s own chips could help lower the cost of AI for its customers. Microsoft has also developed these chips for its own Azure cloud workloads, not to sell to others like Nvidia, AMD, Intel, and Qualcomm all do.

“I look at this more as complementary, not competing with them,” insists Borkar. “We have both Intel and AMD in our cloud compute today, and similarly on AI we are announcing AMD where we already have Nvidia today. These partners are very important to our infrastructure, and we really want to give our customers the choices.”

You may have noticed the Maia 100 and Cobalt 100 naming, which suggests that Microsoft is already designing second-generation versions of these chips. “This is a series, it’s not just 100 and done... but we’re not going to share our roadmaps,” says Borkar. It’s not clear how often Microsoft will deliver new versions of Maia and Cobalt just yet, but given the speed of AI I wouldn’t be surprised to see a Maia 100 successor arrive at a similar pace to Nvidia’s H200 announcement (around 20 months).

The key now will be just how fast Microsoft gets Maia into action to speed up the rollout of its broad AI ambitions, and how these chips will impact pricing for the use of AI cloud services. Microsoft isn’t ready to talk about this new server pricing just yet, but we’ve already seen the company quietly launch its Copilot for Microsoft 365 for a $30-per-month premium per user.

Copilot for Microsoft 365 is limited to only Microsoft’s biggest customers right now, with enterprise users having to commit to at least 300 users to get on the list for its new AI-powered Office assistant. As Microsoft pushes ahead with even more Copilot features this week and a Bing Chat rebranding, Maia could soon help balance the demand for the AI chips that power these new experiences.
https://www.theverge.com/2023/11/15/239 ... astructure

[Spandau Mullet]

https://sharetron.com/@casey/111417613984607584

Casey Kolderup
@[email protected]

I’m so glad CircleCI now has the power of AI to tell me how to fix my test failures

[Naantalin seksuaalimaagikko]

https://sharetron.com/@casey/111417613984607584

Casey Kolderup
@[email protected]

I’m so glad CircleCI now has the power of AI to tell me how to fix my test failures

[Spandau Mullet]

https://arstechnica.com/gadgets/2023/11 ... assistant/

Windows 10 is back, and it’s getting Microsoft’s AI-powered Copilot assistant
But Windows 10's October 2025 end-of-support date is staying put, for now.

In the last two years, Windows 11 has ushered in significant updates for most of Windows' built-in apps, and things like the system tray, Start menu, Settings app, and taskbar have continuously evolved with each new update. But few of these changes have been made available for Windows 10, which is still, by every publicly available metric, the most-used version of Windows on the planet. (Notable exceptions include the redesigned Outlook app and continued development of Microsoft Edge.)

Today, the company is making a major exception: The new AI-powered Windows Copilot feature from Windows 11 is being backported to Windows 10 and will be available in the Windows Insider Release Preview channel for Windows 10. This version of Copilot, which will be branded as a preview at first, will be available for the Home and Pro versions of Windows 10. But it won't be available for the "managed" versions of Windows 10 just yet—Enterprise and Education editions, as well as Pro PCs that are joined to a domain or are otherwise managed by an IT department.

"We are hearing great feedback on Copilot in Windows (in preview) and we want to extend that value to more people," writes Microsoft in a separate blog post. "For this reason, we are revisiting our approach to Windows 10 and will be making additional investments to make sure everyone can get the maximum value from their Windows PC including Copilot in Windows (in preview)."

The Windows 10 version of Copilot will require a PC with 4GB of RAM and at least a 720p display to run; all Windows 11 PCs meet these requirements, so Microsoft didn't need to define these minimums before now. The feature will roll out to North American users and those in "parts of Asia and South America" first and then to other countries "over time."

Microsoft's blog post is a bit light on what Copilot for Windows 10 can actually do. The Windows 11 version can change various system settings and work with documents stored on your PC; the screenshot Microsoft has shared of the Windows 10 version looks mostly focused on Bing Chat-style, ChatGPT-powered text generation.

The Windows 10 November preview update that includes Copilot also makes a few other small changes, including a long list of bug fixes. It also includes Windows 11's "get the latest updates as soon as they're available" toggle in Windows Update, which when enabled will install preview updates automatically rather than making you do it manually.

Though this is the biggest update that Microsoft has released for Windows 10 since Windows 11 replaced it in late 2021, it doesn't change anything about Windows 10's broader support timeline. Version 22H2 remains the last major update available for the OS, and security updates are still scheduled to end on October 14, 2025, less than two years from today. Microsoft representatives avoided answering questions about Windows 10's persistently high usage and the number of PCs that would be unable to upgrade, only saying that Windows 11 adoption was exceeding Microsoft's internal targets.

Though some of those PCs will be able to upgrade to Windows 11 to continue getting security updates, the newer operating system's more stringent system requirements leave a substantial chunk of Windows 10 PCs behind. It's not clear just how many PCs that is, but Windows 11's adoption has slowed in recent months and Windows 10 is still by far the most-used version of the operating system.

Even though Windows 10's end-of-support date is creeping up on us, it's easy to see why Microsoft is extending Copilot support to the older OS. According to Statcounter data for US PCs and devices worldwide, Windows 10 still accounts for over two-thirds of active Windows installs. These users can access some AI functions via the version of Copilot that's built into Microsoft Edge, but Edge only accounts for around 10 percent of all desktop browsing in the US.

Extending Copilot to Windows 10 is a way to get the service in front of many users who otherwise wouldn't access it. And for better or worse, that's been Microsoft's strategy for 2023: casting as wide a net as possible with its AI efforts, getting them in as many of its products as possible as quickly as possible.

Voi vittu Bill Pitääpä heti ensi töikseen googlata miten tämmösen paskeen saa disabloitua siinä vaiheessa, kun moisen käyttöjärjestelmään pakottavat.