Published: October 18, 2025
48
106
1.9k

@SebastienBubeck @MarkSellke this is embarrassing

@demishassabis @SebastienBubeck @MarkSellke getting mogged by Demis for being an utter idiot sandwich is in the top ten AI moments of the year thus far, someone put it in the book and end of year recap, and never let them forget it.

@demishassabis @SebastienBubeck @MarkSellke they've been oneshot by their own creation never had the makings of a nobel laureate

Image in tweet by Demis Hassabis

@demishassabis @SebastienBubeck @MarkSellke The community note though

Image in tweet by Demis Hassabis

@demishassabis @SebastienBubeck @MarkSellke I mean they did literally find the solutions with the help of GPT-5 right? I guess the lack of clarification of how GPT-5 did this can lead to people drawing the wrong conclusions. No doubt useful in accelerating math research tho: @tao/115385022005130505 class="text-blue-500 hover:underline" target="_blank" rel="noopener noreferrer">https://mathstodon.xyz/@tao/11...

@demishassabis @SebastienBubeck @MarkSellke You know you are caught being stupid when THE demis calls you embarrassing💀💀💀💀

@demishassabis @SebastienBubeck @MarkSellke I am mixed. Sebastien did say "found" as opposed to "discovered", but it definitely feels like it is meant to sound like the latter. Some @OpenAI definitely need to cool it with the "scientific breakthrough" hype train as it ends up causing loss of trust in the entire space.

@demishassabis @SebastienBubeck @MarkSellke "dad, where were you when Demis Hassabis called out a senior researcher at OpenAI for hypebeasting GPT5?" well son...

@demishassabis @SebastienBubeck @MarkSellke Just post your Nobel prize and that says enuf

@demishassabis @SebastienBubeck @MarkSellke GPT-5 is great at Googling 😏

@demishassabis @SebastienBubeck @MarkSellke Speaking of embarrassing, found this academic scam masterpiece next to my paper (and also yours) in main Nature https://www.nature.com/article... Embarrassingly slow and stupid editorial board is allegedly doing some external validation lol. I believe little of your attention would help

@demishassabis @SebastienBubeck @MarkSellke Such a shame to see OAI turn into a hype machine. Not great for the industry. Don’t hype, just ship.

@demishassabis @SebastienBubeck @MarkSellke So misleading and dishonest. Also if I ever get ratioed, at least let it be @demishassabis to do it. Or @greg16676935420. I could live with Greg ratioing me.

@demishassabis @SebastienBubeck @MarkSellke Wow I have never seen Sir Demis to be a reply guy, forget about some random OpenAI researcher, wonder what happened.

@demishassabis @SebastienBubeck @MarkSellke Huh? It says he found unknown references thru search and linked them up? Which is true?

@demishassabis @SebastienBubeck @MarkSellke Didnt he mention "found" though and not discovered🤔

@demishassabis @SebastienBubeck @MarkSellke They should be embarrassed.

@demishassabis @SebastienBubeck @MarkSellke I muted "OpenAI", they are a sh*t company with fake researchers.

Share this thread

Read on Twitter

View original thread

Navigate thread

1/33