Gonna be great watching critical infrastructure collapse as a result of using this

[kevin@doyle.boston]

@virtualbri @vmstan @tomshardware the AI thought rather than fix the code, letting the nuclear power plant meltdown was the more practical option

[ottaross@mastodon.social]

@virtualbri Slop code that few people have the skills to refactor or debug. Sounds like a recipe for disaster… coming soon to your bank.

[raganwald@social.bau-ha.us]

@virtualbri Matt Levine has—as usual—a great take on this.

You know how Musk makes some wild promise or demo like a single car delivering itself via carefully pre-mapped route with chase cars offscreen, gets a stock pump, and then moves on?

Anthropic can do a different thing: Short a public company and then announce a demo of how Claude or whatever is going to completely disrupt that corner of the SaaS and Consulting ecosystem.

Instant profit, nearly zero risk

[craigduncan@mastodon.au]

@raganwald @virtualbri

Its like blowng a smaller bubble inside a larger one.

[artair@ohbear.wtf]

@virtualbri @tomshardware I'm still wondering how all this looks once the bubble pops. And I can't imagine that staid, risk-averse places like banks will rush to embrace this... especially since banks hold so much AI debt at risk of default right now.

"Should we really swap what works with experimental stuff when these vendors can't even pay the interest on their debts?”

IBM CEO warns that ongoing trillion-dollar AI data center buildout is unsustainable — says there is 'no way' that infrastructure costs can turn a profit

Krishna’s cost model challenges the economics behind multi-gigawatt AI campuses.

Tom's Hardware (www.tomshardware.com)

[Guest]

@virtualbri
business people: ibm's stock crashed because people are going to use claude to write cobol! (positive)
reality: ibm's stock crashed because people are going to use claude to write cobol (negative)

[wcbdata@vis.social]

@virtualbri IBM recently recognized that AI was killing its pipeline of qualified engineers and simultaneously killing the quality of its codebase, so it scaled back its AI investments and dramatically ramped up hiring of junior engineers/developers. If I had disposable income, I might consider buying IBM on the dip (as the cool kids say).

[larrymodell@sfba.social]

@virtualbri There was a time when corporations refused to use open source software because there had to be somebody (a vendor) to sue if things went terribly wrong.

The thing is, some human will have to sign off on that AI-generated code (or legal brief, or engineering design, etc.) They will have to certify that they have reviewed it, understood its "logic," and will stand by its outcomes, at the risk of being fired, sued, or prosecuted.

Kids, I think we've found an AI-proof career for you: fall guy/patsy for AI catastrophes.

[troy@opencoaster.net]

@virtualbri @tomshardware sad thing is there’s some older devs fully embracing it. Code aside, I fear the worst part will be the loss of knowledge and context of those systems. Doubt there’s enough docs.

[virtualbri@mastodon.online]

@artair @tomshardware I DESPERATELY WANT media and banks to actually look into these things and *analyze* properly to separate nonsense from reality. It's a disservice to everyone to not evaluate these technologies and just repeat PR drivel and to feed billions of dollars into it that could go to actually valuable businesses who will now never get financed.

[monsieur_lepetit@toot.community]

@virtualbri
"HAL, open the door!"
"I'm sorry Dave, I don't know how"

[virtualbri@mastodon.online]

@larrymodell Yeah it's crazy. I don't get how with the errors that are already cropping up that companies haven't been sued. Like medical mistakes already happening.

[javensbukan@mstdn.ca]

@virtualbri Uhhh, could we not use AI to write things that run critical infrastructure. K thx.