Conversation

Replying to
So for example, I react to the potential for traffic delays by getting to the airport extra early, but that in turn is driven by my sense that I hate sprinting for the gate, and extra-hate the energy demands of rearranging plans if I miss a flight. I do not like having to spike.
18
"Minimize energy" is a very illegible optimization problem, since our energy efficiency is a complex function of physiology, cognitive style, and output efficiency. I'm more wasteful in sprint/spike mode. Otoh, I both hate planning and am inefficient/chafing in working to one.
1
17
So I end up solving for minimal energy use/least resistance by a) adopting a highly improv style b) building in lots of time for everything, so I can figure things out by trial and error in a *relaxed* unhurried way. I like to iterate, but not fast.
26
A preference for improv over planning is about narrowing the uncertainty band (improv uses more up-to-date info) but also about limiting rework energy (since waterfall plans need higher-energy reworking when they fail, where improv is typically just 1-step backtrack)
1
27
Replying to
Hmm, I am very skeptical about this article. Seems to me a deliberately uncharitable reading. Could we apply the same critique ("isn't sufficient to build GAI") to Friston's FEP? Similar aims: a theoretical foundation of agency. Maybe I'm wrong, have you seen other critiques?
1
Replying to
Gotcha. If you care to share any more details about the substance of the critiques, I'd be very interested -- might save me from going down the wrong rabbit hole!
1
1