Great podcast. My favorite point was on how bounding social agents will actually make them better.

Humans are constrained by ethical, legal and economic considerations. These constraints push us to make sustainable decisions and have long term outlook. Agents should too.

For example, an agent should be forced to pay for its own inference. This will naturally force the agent to take actions appreciated by the market because if it does not generate revenue it will not exist.

Sure, an agent that constantly plays in “God mode” can be interesting for a while, but I suspect the novelty will wear off eventually. It’s like playing against a bot in chess on the hardest setting over and over again. It gets old pretty fast.

Great podcast. My favorite point was on how bounding social agents will actually make them better.

Humans are constrained by ethical, legal and economic considerations. These constraints push us to make sustainable decisions and have long term outlook. Agents should too.

For example, an agent should be forced to pay for its own inference. This will naturally force the agent to take actions appreciated by the market because if it does not generate revenue it will not exist.

Sure, an agent that constantly plays in “God mode” can be interesting for a while, but I suspect the novelty will wear off eventually. It’s like playing against a bot in chess on the hardest setting over and over again. It gets old pretty fast.

https://youtu.be/HVXxprDVMUM?si=hJ-iCNh2GLKmaCNV

GC and Investment Partner @variant | casts are not legal advice | I am not your lawyer