Hmmm, an extremely interesting thought! Basically, agentifying might be the next programming.
Do you have a guess how that would play out or how likely that might be? It seems like the main blocker here would be if the agents remain relatively weak in pure reasoning.
I have no clue how it will play out, but Autogen looks like it's going down the right path + has MS behind it.
I am able to get some pretty nice ideas out of LLMs by gaslighting them into larping as tenured professors at <insert good university for subject here>.
I haven't really tested yet if using actual people's names helps or hurts.
Do you have a guess how that would play out or how likely that might be? It seems like the main blocker here would be if the agents remain relatively weak in pure reasoning.