Microsoft Makes Multibillion Dollar Investment in OpenAI

[Tsing]

OpenAI is the AI research and deployment company behind ChatGPT and DALL·E 2, two AI projects that have gained popularity for allowing users to generate realistic conversations and art, respectively, by simply typing in text. Microsoft announced today that it is extending its partnership with OpenAI with a new multiyear, multibillion dollar investment, something that will go toward accelerating the company's AI breakthroughs for supercomputing and other projects. This marks the third phase of Microsoft's long-term partnership with OpenAI, which already include the investments that the Windows maker made in 2019 and 2021.

See full article...

 

[Elf_Boy]

I have a friend with a sub to this or a similar service.

The quality of the art is surprisingly good. Perhaps scarily good.

He did say something about it not working well on AMD gpu’s and had to get the run it on their servers option. Guess it uses cuda?

 

[Uvilla]

I have a friend with a sub to this or a similar service.

The quality of the art is surprisingly good. Perhaps scarily good.

He did say something about it not working well on AMD gpu’s and had to get the run it on their servers option. Guess it uses cuda?

Didn't know you can locally run it at all. Ive used dall e 2, and the text one (not the one with computer language). Its fun.

 

[Burticus]

Multi billion dollar investment in AI. But still laying off 15k people? Checks out.

 

[Grimlakin]

Didn't know you can locally run it at all. Ive used dall e 2, and the text one (not the one with computer language). Its fun.

Keep in mind that organizations need to be able to teach the ai but also control their own data. Especially if a chat bot as example is involved in any HIPPA data conversations. So the ability to run it on standalone hardware (ish) is important.

Being able to ingest data from specific sources via pulling and polling of external data and merge that with internally generated data streams while not exposing or exporting ant of the internal data externally is important to a good ai setup that enterprise organizations can implement securely.

The cost of the hardware to develop this ai and run it over the cost of manpower to staff for the same tasks is the balancing act. If an ai farm can reliably handle hundreds of chat with human like informed interactions before needing actual human engagement that would be ideal.

The issue is then with customers that specifically want to opt out of AI interaction. For their own moral or privacy concerns.

Would you like to chat with (company name) ai now about your issue or would you rater wait xx minutes for a live agent?