On a hiking trip once, three decades ago, I accidentally coined (but failed to trademark) the phrase “self-modern, post-parody”.
I think I have finally found a use for it.
Want another example? Here, fully modern and way beyond parody, is Sam Altman — whose company, mind you, has never turned a profit — promising to use AI to capture the value of a large fraction of the entire universe (the earth’s light cone, to be exact).
It’s almost as if the AI community wants to satire itself.
Or how about this?
And when it is not satirizing itself, big tech is regularly and predictably falling on its face, the same way, over and over again, with ever larger amounts of money.
Funny name – Behemoth – for absolutely massive model that so far isn’t working. Wikipedia helpfully traces the term to a “primeval chaos-monster” in the Book of Job. What could possibly go wrong?
Speaking of chaos monsters, Elon promised us that Grok 3 would be genius level:
But why pick on Elon in particular?
Here’s OpenAI’s o3 (you know, the one that Tyler Cowen confidently declared was AGI), with problems of its own:
Consumers are starting to catch on:
For some more serious AI discussion, check out three truly excellent podcasts I appeared on this week:
On scaling, neursymbolic AI, sentience and more in The Most Interesting Thing in AI and other topics, with the unique, amazing, and indefatigable Nick Thompson, CEO of The Atlantic and subscriber to this very Substack, fresh off a world-beating 50 mile run he had raced in (and won) the day before.
On politics and AI, and how they have become deeply intertwined, with the fearless and widely followed
brothers.
A full-on debate over “AI Doom” with Liron Shapiro, in which I was ambushed (in a funny way) with my own obituary – and in which I argued for the more optimistic view on AI (viz., that even in the worst case it probably won’t lead to every single human on the planet dying). Tune in for some Gary optimism!