We spoke to a Stanford prof on the tech and social impact of AI's powerful, emerging 'foundation models'

Even if you haven't heard of "foundation models" in AI, you've probably encountered one or more of them in some way. They could be rather pivotal for the future of not only machine learning and computer science but also society as a whole. Foundation models are called this because they are the base upon which myriad …

[Filippo]

That was very interesting, thanks!

[nautica]

Kill it. Kill it NOW!

From the very first paragraph comes the reason to kill this brand-new, glitzy, Gen-Z whatever-it-is deader than the proverbial door-nail:

"...They could be rather pivotal for the future of not only machine learning and computer science but also society as a whole..."

Just exactly and precisely like Facebook.

[Throatwarbler Mangrove]

Re: Kill it. Kill it NOW!

Can't get the genie back in the bottle, I'm afraid. Wintermute says no.

[Anonymous Coward]

How 'foundarion modules' are developed

From the omniscient XKCD:

https://xkcd.com/2494/

[terrythetech]

Who decides?

"We can have constraints on what this technology can be used for."

[nautica]

Re: Who decides?

"Who decides?"

Absolutely correct; and elegantly simple observation, and question.

And, furthermore,...

who decides objectively, and who enforces, with the appropriate forceful consequences, the decisions?

[Short Fat Bald Hairy Man]

Re: Who decides?

Looking at the Tinnit Gebru case, will anybody be allowed to do it? Not within the companies.

And outside will be even trickier I think.

Looks like the highway to hell.

[Paul Hovnanian]

Re: Who decides?

So then someone is in charge of periodically checking NPM to make sure a SkyNet.js package hasn't been released?

By the time the 'who' has been determined and a suitable set of rules established, the horse will have left the barn.

[nautica]

Re: Who decides?

"...By the time the 'who' has been determined and a suitable set of rules established, the horse will have left the barn."

Quite right.

Certain individuals have recognized that this is a major failing of our otherwise wonderful 'information age'; they take advantage of the fact that the societal-legal means of restriction are woefully slow--and sometimes, downright inadequate--to respond in a meaningful way; a very rapid way--and can be made a mockery of by people whose modus operandi is "move fast and break things". Unfortunately, it is becoming quite clear that even this MO needs to now be updated to "move fast, break things, and kill people".

[Geez Money]

Re: Who decides?

The problem is simpler than that. The notion that we can pass some laws to prevent research into an area is wrongheaded to begin with. 1) good luck enforcing a law that can be violated with an offline chromebook (now that the model is computed) and 2) good luck getting every country on Earth to agree with you rather than taking you not researching this area as a strategic advantage.