It might still be a little slow (I'm not sure if the 16 seconds to compute an action is fast enough for commercial use cases), but this is definitely exciting and seems like a great step forward.
Like others have said, its an interesting avenue for AGI. The joint embeddings would be closer to thinking than the current LLM token work. LLMs look like they have a lot limitations for AGI (although who knows if we have another crazy scale up? but that extra scale is looking difficult right now).
For those that use LLMs, how many still use a subscription? Is it worth it? I've found a counsel of LLMs on free tier to be quite effective for my needs, but I'm curious if anyone feels strongly about needing to pay to get the advanced models.
While I tend to agree, I wonder if synthetic data might be reaching a new high with concepts like Google's AlphaEvolve. It doesn't cover everything, but at least in verifiable concepts, I could see it produce more valuable training data. It's a little unclear to me where AGI will come from (LLMs? EBMs - @LeCun)? Something completely different?)
I know this is the terminology, but I'd argue that the activations are the actual thinking. It's probably too late to change that, but I wish people would refer to thinking as the work Anthropic and Deepmind are doing with their mech interp