Greg Kamradt(@GregKamradt):> Even if you had Noam Shazeer level models tomorrow they'd still be experiment bottlenecked, so it changes the shape of takeoff. very important factor to model. I think we under estimate the amount of evidence (in the deutsch sense) that we have already The trite saying you’ll hear is, “there are $B companies on pubmed, you just need to go find them.” This applies to all of the signal we have from *current* experiments. There is so much signal coming our way that the blocker is intelligence, not necessarily more signal. AGI with the hadron collider as a tool call away will be able to do a lot… Now of course this is on the immediate 5-10yr timelines. Sooner or later we’ll exhaust our evidence and need new sources of it.

2026.08.17 19:56

> Even if you had Noam Shazeer level models tomorrow they'd still be experiment bottlenecked, so it changes the shape of takeoff. very important factor to model. I think we under estimate the amount of evidence (in the deutsch sense) that we have already The trite saying you’ll hear is, “there are $B companies on pubmed, you just need to go find them.” This applies to all of the signal we have from *current* experiments. There is so much signal coming our way that the blocker is intelligence, not necessarily more signal. AGI with the hadron collider as a tool call away will be able to do a lot… Now of course this is on the immediate 5-10yr timelines. Sooner or later we’ll exhaust our evidence and need new sources of it.

Sholto Douglas@_sholtodouglas

2026.08.15 17:55

>may be really civilizationally useful in giving us time to adapt to what might be a fast takeoff. Agree - this has really interesting implications on RSI, which is that even if you had Noam Shazeer level models tomorrow they'd still be experiment bottlenecked, so it changes the shape of takeoff. very important factor to model. >But my argument would be the same: rather than worrying about inequality we should focus on growing the pie so fast that even the bottom of the income/compute distribution are much better off here in America than almost anywhere else. 1000% agree. My median expectation is that compute spend follows the 2-3x yoy trendline (so that next year it is >2T), and that a given level of intelligence keeps getting 10x cheaper every year due to efficiency improvements throughout the stack which will hopefully compound into intelligence too cheap to meter. Will take you up on that dinner in Boston sometime, or let me know next time you're in SF! And don't worry - I drink, can't take the Australian out of the guy

Forward to community