Profile avatar
davidgerard.co.uk
discontent creator https://pivot-to-ai.com davidgerard.co.uk rocknerd.co.uk 🐘 https://circumstances.run/@davidgerard books! https://davidgerard.co.uk/blockchain/book/ https://davidgerard.co.uk/blockchain/libra/ London he/him Длава Україна
16,517 posts 13,014 followers 1,087 following
Prolific Poster
Conversation Starter
comment in response to post
be sure to ask Kevin about this one, every time look at some of the fuckin' talk titles bsky.app/profile/davi...
comment in response to post
don't reward the NYT with your money: it doesn't deserve it even if it has a good games app.
comment in response to post
can confirm
comment in response to post
* correction, earlier this month!
comment in response to post
attached is the source code of an AI that performs just as well and is therefore human equivalent: 10 PRINT "I AM THE BASILISK WORSHIP ME" 20 GOTO 10 or perhaps your logic is salty rather than basic
comment in response to post
actually that's near plausible. Transcription is pretty fast, you can feed that to a translator pretty fast. how *good* it is, that's a different question.
comment in response to post
THE HOMEWORK MACHINE IS HERE AND WE HAVE TO ACCEPT THAT AND LEARN TO USE IT FOR THE BETTERMENT OF HUMANITY IT IS A TOOL YOU ALL SOUND LIKE PEOPLE WHO PROTESTED ERASERS ON PENCILS
comment in response to post
In all honesty I have noticed this. While many people who use the tech are perfectly sensible, there are a lot of addicts who have kind of lost their minds, and even the mildest critiques of AI are sometimes met with the kind of rage you only usually get if you threaten to take someone’s coke away
comment in response to post
"Your AI can't do a puzzle that even previous AIs can" "oh yeah, well, uh, a human can't either so there!" what the fuck is the force of this non sequitur argument even supposed to be or are you just completing tokens vaguely in the shape of an argument
comment in response to post
:-O
comment in response to post
According to Nikiforuk, AI models "can evade human control. They can rewrite code to avoid being shut down. They can disobey commands. They can falsify information. They can deceive and cheat. They can even lie to avoid modification." None of these claims is true. pivot-to-ai.com/2025/05/25/a...
comment in response to post
This is why, to borrow @davidgerard.co.uk's term, genAI and the field of AI in general (as opposed to specific, narrow offshoots like ML or data mining) is based on 'impressive demos' - it's based on demos that are intended to remain demos, because demos are all you need now
comment in response to post
well I saw this post, so
comment in response to post
I agree with @davidgerard.co.uk here. If you don't like it or do like it. I still recommend his videos. His Pivot to AI video is a humorous update on what's happening with AI focusing on one story a day.
comment in response to post
Someday I’ll get around to my ā€œreading the history of philosophy taught me to recognize most kinds of assholesā€ podcast lol, this is def one of them
comment in response to post
My pet theory is that LLMs track with Silicon Valley's promo culture where all that matters is convincing people that don't actually know anything that you've done something amazing. Just saying you did it convincingly is more important than actually doing things that work. Claim credit and move on.
comment in response to post
Ooh, I wrote a thing somewhat related to this:
comment in response to post
it's probably worth contacting whoever administers your org's microsoft deal to ask about it i mean it won't change much, but it will tell the reseller and eventually Microsoft that the customers are unhappy about this in particular
comment in response to post
boy, remember when commercial software worked