>Will there be any Chinese Fable-level models by March 2027, defined as a model with an ECI score of 160 or higher? We give this a 73% (65% to 80%) chance.
In April, you all wrote "When will there be an open source AI with Anthropic Mythos-level capabilities? Forecasters’ on average expect it to arrive on December 15th 2026, with an aggregate 90%-ile range of May 17, 2026 to May 29, 2028"
I'd love to understand the difference between these two predictions. Was the "Mythos-level" threshold set lower? Or have forecasters shifted their estimates back?
These seem pretty consistent. Mythos would be at a similar level than Fable right now. I guess we have a bit more information about where Chinese models are now. I've also wondered more about whether China can physically get the chips & whether that will be a bottleneck; just wrote this thread: <https://x.com/NunoSempere/status/2076738479262945318> as a result of your question.
Like, I think the answer is that by default, if lines continue, they would reach those capabilities, but there is a physical bottleneck here which is having enough chips, and that's what could potentially stop the line from going up.
Thanks for the replies! Found that interesting. And now reading through it a second time, I realize that the two predictions are asking different things at it’s implicitly 73% at March 2027, not 50%. Which makes more sense.
As a side note, it might be interesting to have an agent run through your back catalogue to find comparisons like these (even if they’re slightly different each time). For instance, you all have made a good number of predictions about Ebola — this might give a sense of how our information has updated over time.
>Will there be any Chinese Fable-level models by March 2027, defined as a model with an ECI score of 160 or higher? We give this a 73% (65% to 80%) chance.
In April, you all wrote "When will there be an open source AI with Anthropic Mythos-level capabilities? Forecasters’ on average expect it to arrive on December 15th 2026, with an aggregate 90%-ile range of May 17, 2026 to May 29, 2028"
I'd love to understand the difference between these two predictions. Was the "Mythos-level" threshold set lower? Or have forecasters shifted their estimates back?
Thanks as always!
These seem pretty consistent. Mythos would be at a similar level than Fable right now. I guess we have a bit more information about where Chinese models are now. I've also wondered more about whether China can physically get the chips & whether that will be a bottleneck; just wrote this thread: <https://x.com/NunoSempere/status/2076738479262945318> as a result of your question.
Like, I think the answer is that by default, if lines continue, they would reach those capabilities, but there is a physical bottleneck here which is having enough chips, and that's what could potentially stop the line from going up.
Thanks for the replies! Found that interesting. And now reading through it a second time, I realize that the two predictions are asking different things at it’s implicitly 73% at March 2027, not 50%. Which makes more sense.
As a side note, it might be interesting to have an agent run through your back catalogue to find comparisons like these (even if they’re slightly different each time). For instance, you all have made a good number of predictions about Ebola — this might give a sense of how our information has updated over time.
But likely I’m just being lazy.