In a demonstration at the UK’s AI safety summit, a bot used made-up insider information to make an “illegal” purchase of stocks without telling the firm....
I work on chatbots for a big tech company. Every team is trying to use GenAI for everything. 90% of the stuff they try won’t work. I have to explain that LLMs can’t actually think at least three times a week. The hype train was too strong. Even calling it AI feels misleading.
That said, there are some genuinely great applications for LLMs that i’ve enjoyed looking into.
They don’t reason, they’re stochastic parrots. Their internal mechanisms are well understood, no idea where you got the notion that the folks building these don’t know how they work. It can be hard to predict/understand how an LLM generated a given prompt because of the huge training corpus and statistical nature of neural nets in general.
LLMs work the same as any other net, just with massive sample sets. They have no reasoning capabilities of any kind. We are naturally inclined to ascribe humanlike thought processes to them because they produce human-sounding outputs.
If you would like the perspective of real scientists instead of a “tech-bro” like me I would recommend Emily Bender and Timnit Gebru. I’d recommend them as experts without a vested interest in the massively overblown hype about what LLMs are actually capable of.
That may be true for warehouse employees, but the corporate offices are a toxic mess of shitty culture and dated ideas. I’ve never seen a tech department bleed so much underpaid talent to Amazon.
When I quit because they tried to force me back into the office mid-pandemic (August 2020) I had multiple offers for fully remote positions with twice the salary within a few weeks.
But yeah, if you are a cashier at a warehouse or whatever I hear it’s a solid gig.
It’s not actually about listing the fees. They’re worried that if they have to list the fees, customers will realize they’re paying 19.99 a month to rent a router, or are getting charged for a land line they didn’t ask for.
When I used to do copywriting for junk SEO, I began to suspect that my editor didn’t actually read anything I wrote and just passed it through a content uniquness filter, so I started putting in random references to HP Lovecraft stories in the articles I got assigned.
They all got published, no questions asked. For a while if you searched “Homeopathy and the Esoteric Cult of Dagon” my content was the only result
cabbage (feddit.de)
New machine day (lemmy.world)
Profitec pro 500 PID paired with an ECM S-Automatik 64....
AI bot capable of insider trading and lying, say researchers (www.bbc.com)
In a demonstration at the UK’s AI safety summit, a bot used made-up insider information to make an “illegal” purchase of stocks without telling the firm....
Israel calls for all 1.1 million civilians to leave Gaza City within 24 hours (www.reuters.com)
When the pizza party is too expensive, you go with the EncourageMint (lemmy.ca)
deleted_by_moderator
FCC says “too bad” to ISPs complaining that listing every fee is too hard (arstechnica.com)
Aww … poor little ISPs.
Starfield is Bethesda's Least Buggiest Game to Date, Say Sources (insider-gaming.com)
Biden calls China a 'ticking time bomb' due to economic troubles (ground.news)
What would be an eleventy-first birthday in dog years? (lemmy.ca)