OMG!! THIS is what Trump and Musk are up to!! (Forget the sensational distractions!)

[highplainsdem]

71. More than one tech bro is involved, and journalist Gil Duran is doing the best writing about this.

A number of threads about his work have been posted here, including this

https://www.democraticunderground.com/100219994149

and this today:

https://www.democraticunderground.com/100220006526

Conrad Goehausen doesn't seem to have any background that would give him any expertise or connections. He's making guesses, based on what he's reading that others have written.

As for how well Musk's Colossus works - it's just another ridiculously expensive and environmentally damaging piece of LLM crap that hallucinates, judging by his Grok AI and how badly it works, and it will be very likely to mess up any government program it's used to run.

https://www.businessinsider.com/elon-musk-xai-colossus-supercomputer-nvidia-reid-hoffman-chris-lattner-2024-9

Not that Musk cares about that, when his apparent goal is to cripple the US government.

But so far the AI he's showing off on Twitter excels at confusing the identities of people, which will mean endless problems when used to analyze vast amounts of data about hundreds of millions of people.

Musk has used a name taken from a science fiction film - https://en.wikipedia.org/wiki/Colossus:_The_Forbin_Project - about a computer becoming sentient to make his AI seem more impressive and scary. But LLMs will never become sentient.

Something Gary Marcus has been pointing out for some time, and Yann LeCun is finally agreeing with him:

https://www.msn.com/en-us/money/other/nobody-in-their-right-mind-will-use-genai-llms-in-the-next-5-years-meta-chief-ai-scientist-yann-lecun/ar-AA1xMgcV

Musk is just spending more than anyone else running toward an AI dead end. But at this point there's nothing to indicate his AI is better, or even as good as, other AIs.

And none of them are trustworthy enough to run a government on.