[pkoiralap]

This is fascinating research. I feel this is a significant leap in interpretability research. Since we know J-Space exists and is bi-directional, we can train models on the same and come up with meta cognition abilities.

I also fear that the big corporations might use the same to run targeted ads, capitalistic shenanigans. Which they might already be doing through system prompts.

[marshray]

Such an inspection capability might also be used to target ads to LLMs, which would then be more likely to mention or recommend those products and services.

[motoxpro]

super interesting

[tclancy]

The prior art here is “Have Fun, Good Luck, Don’t Die”.