Conversation Designer, Sci-Fi Enthusiast, Sometimes Author | @ Melbourne, Australia
3 followers 2 following 15 posts
view profile on Bluesky
I joined Facebook shortly after Chamath left. When he became a podcast/VC guy, I asked an engineering director on my team who’d reported to him what he was like. He thought for a second. “I want to try to be nice, you know,” he said. “Chamath was one of the dumbest motherfuckers I’ve ever met.”
FT is a fairly sober, reputed financial publication, so panic and worry is not something they'll ever do. The fact they published this, even as an op-ed, should worry us.
If "pretty close" is a compromise between reality and what Altman wants to hear, OpenAI is fucked beyond belief.
Quick feedback for you: Finish the article before replying :)
Brother, scroll down.
It's ok that you didn't read the article, but it's deeply embarrassing how confidently you attempted to "take it down." The article literally acknowledges what you said AND THEN SHARES OPENAI'S OWN DATA SHOWING THE SAME THING. 🤡
Whenever I see an AI company showing off their best AI use-cases.
I work at one of these companies. I worked on Gemini + I work at another AI player. @edzitron.com knows more about AI than Newton and Roose. He challenges AI companies' claims rather than regurgitating press releases and whatever Sales 'Engineers' tell them over $22 cocktails at a fancy bar in NYC.
Oh ok
Ok and?
What do you mean a generic media generator made by the biggest corporations on earth that are run by conservative bootlicker isn’t punk?
A secondary unto itself isn’t really meaningful. But a company with a single-digit conversion rate looking at a near 50% jump in valuation? Yeah what.
In an apparent race for the future, why would *any* AI-focused companies laying people off? Meta, Microsoft, Google — surely, if AI is an existential threat it'd be all-hands on deck, right? And yes, they did layoff people useful in the 'war': engineers, content designers, product designers, PMs.
There's no way even 50 geniuses working 24/7 could create a "fine-tuned" model (so are they tuning an established model?) that beats Sonnet 4, GPT5 or G2.5+. They're still tying their shoes after the race has already begun. Source: I worked on training and evals at Google (Bard->Gemini).
A lot of folks are dunking on you, and it's unfair. Their point is valid: drug dealers don't give away free stuff to get you hooked. The deeper point is that AI products aren't sticky. ChatGPT boasts 500m weekly actives... but only 15m subscribe. That's not what a sticky product looks like