Pascal Hary В LinkedIn: Microsoft Plans To Build Openai, Chatgpt Features Into All Products, Chatgpt is very expensive, but Microsoft has a plan to save – Numerama
Chatgpt is very expensive, but Microsoft has a plan to save
You think that AI is an innovation motor at work and that it can release the potential of our professions ? Microsoft France launches at Station F Le . An unprecedented program, aimed at accelerating 12 startups and French software publishers who wish to reinvent today’s trades thanks to the generative AI. Supported companies may benefit from including technical resources, access to the most innovative AI models of the moment (GPT-3.5, GPT-4) and a workspace dedicated to Station F. The program for a duration of will be punctuated between technical workshops common to 12 selected companies and a on the part of Microsoft France and the program partners. We are looking for B2B post MVP companies that have identified first IA boxes for their solution and which will be able to take full advantage of Microsoft technological bricks ! Applications are open until October 8 https: // lnkd.In/Erdi683y #GenaisTartUpprogramm #ai
Публикация уастника Pascal Hary
You think that AI is an innovation motor at work and that it can release the potential of our professions ? Microsoft France launches at Station F Le . An unprecedented program, aimed at accelerating 12 startups and French software publishers who wish to reinvent today’s trades thanks to the generative AI. Supported companies may benefit from including technical resources, access to the most innovative AI models of the moment (GPT-3.5, GPT-4) and a workspace dedicated to Station F. The program for a duration of will be punctuated between technical workshops common to 12 selected companies and a on the part of Microsoft France and the program partners. We are looking for B2B post MVP companies that have identified first IA boxes for their solution and which will be able to take full advantage of Microsoft technological bricks ! Applications are open until October 8 https: // lnkd.In/Erdi683y #GenaisTartUpprogramm #ai
- Копировать
Чтобы Просмотреть или add a how, Выполните Вход
Another Exciting Day for the Microsoft Community As We Further Advance Our Ai Offerings With Microsoft Copilot and Newly Unified Features in Some of our Most Used Tools and Applications, Like Microsoft 365 COPILOT, Edge Web BROWSER, AND THE NEW BING. I encourage Everyone to check out Yusuf’s Blog post to read How We’re Continuing to Drive Digital Transformation Through Our Enhanced Ai Tools. And huge congratulations to all involved in today’s event and Bringing thesis Incredible Capability to Life. https: // lnkd.in/gfypmfm
Announcing Microsoft Copilot, Your Everyday Ai Companion – The Official Microsoft Blog
https: // blogs.microsoft.com
- Копировать
Чтобы Просмотреть или add a how, Выполните Вход
Heirloom’s Direct Air Capture Hubs Are Too. TODAY WE ANNOUNCED One of the Large Carbon Removal Deals with Heirloom. This Agreement Will Remove Up to 315k Metric Tons of Co2 and Provide Project Finance to Build Future Dac Facilités. https: // lnkd.in/gaig4yhk #msftadvocate
WSJ News Exclusive | Microsoft will use Carbon-Absorbing Rocks to Meet Climate Goals
wsj.com
- Копировать
Чтобы Просмотреть или add a how, Выполните Вход
Groupama incorporates the generative #ia within its organization with two key principles: A #Innovation always at the service of its customers, a responsible and ethical approach at the heart of reflections and the implementation of use cases. Discover the backstage of this project, the keys to success, the results and the future prospects during a session focused return to Big Data & AI Paris https: // lnkd.in/eir9zhyc #Microsoftfrance #Azure
- Копировать
Чтобы Просмотреть или add a how, Выполните Вход
Our new space adventure game, starfield, is a “sprawling celestial journey” that has the potential to become a “crowning jewel” in the game pass portfolio. #msftadvocate #gaming #xbox https: // lnkd.In/DKUW73XT
Starfield’s 1,000 May be one giant leap for game design
https: // www.nytimes.com
- Копировать
Чтобы Просмотреть или add a how, Выполните Вход
Chatgpt is very expensive, but Microsoft has a plan to save
Microsoft has a project consisting in developing your own chips, optimized for artificial intelligence (AI). Components that should allow him to spend less money to run his projects in AI, such as Chatgpt.
We saw it with the partnership with several billion dollars concluded with Openai: Microsoft puts the package in artificial intelligence. For example, the American company has integrated the latest version of Chatgpt (with GPT-4) in Bing. She also unveiled Copilot. She also added an image generation service on the fly, with Dall-E.
Microsoft has, in a way, no choice: its rivals (Google, Facebook and Amazon) are also launched in the AI race. Not being in the blow would exhibit the Redmond firm a risk of downgrading. If it had nothing to offer, its customers could turn to competition, which would be happy to welcome it. It would be a disaster.
But there are also strategic considerations. Maybe with AI, Microsoft has a golden opportunity to weaken Google, including in online research, a sector where the Mountain View firm has dominated all the debates for twenty years. This is a real revolution that is looming and Google knows that its future is threatened by AI and that worries it.
Running the AI is expensive. Extremely expensive, even. In February, Forbes headlined ” Chatgpt burns millions of dollars every day »». An amount reassessed in April by an analyst, who issues a lower estimate. Daily expenditure to make the conversational agent work (chatbot) would rather be around $ 700,000. It remains very expensive.
This is not surprising, however: AI requires colossal computer resources for those who want to launch services and products whose scope is global. Means all the more important when it is addressed to millions of users around the world, and who all expect instant results or almost.
This power is necessary for the constitution of the models on which algorithms will be set, but also for the entire training sequence itself (automatic learning). To this is then added the execution of the requests of Internet users, which can be millions, and who also massively solicit resources.
Microsoft’s Athena project for fleas dedicated to AI
Under these conditions, it is not surprising to learn that Microsoft is working on the development of his own chips. In its April 18 edition, The Information announces that Microsoft is preparing a processor to contain the increase in automatic learning costs. This project has a code name, Athena, and has existed since 2019, according to the American media.
This date, which confirms two sources consulted by our colleagues, turns out to be also the year of the first substantial investment of Microsoft in Openai. At the time, the amount was a billion dollars. A sum that appears small today, since the software giant has multiplied this bet by ten, four years later.
According to The Information, the first chips are available for Microsoft and Openai, to test them. The objective of the two partners is to find a way to reduce the manufacturing costs of these components, crucial, and also to develop models that require less electrical consumption. Because AI also needs a lot of energy.
Which is true for Openai (which built Chatgpt) and Microsoft is just as much for the other companies launched in this competition. Besides, the great rivals of American tech also design their own chips. This is the case with Google, Amazon and Facebook. These groups then entrust the flea industry to produce them.
The development of more efficient chips, less expensive to develop and having a controlled, even lower energy imprint appears as one of the major issues to be resolved for the future of AI development. In this area, fundamental research turns out to be crucial, in parallel with the innovations that take place in the flea industry.
In this area, advances exist and France has results. At the beginning of March for example, the National Center for Scientific Research (CNRS) pointed out the work of a team associating CNRS laboratories, the CEA Leti (Laboratory of Electronics of Information Technologies of the Atomic Energy Commission) and the Hawai startup.tech.
Their research has made it possible to build a prototype machine that performs a task of AI (recognize a human gesture) using thousands of times less energy than a traditional solution. Here, a new technology of nanocompains, the Memristors, has been used to reduce data movements. Typically the kind of progress that will interest the AI sector.
The future of Numerama is coming soon ! But before that, we need you. You have 3 minutes ? Answer our investigation