Thus, in the H-JEPA framework, this could potentially be implemented as a host of separate components in the Critic module, trained to evaluate whether the plans inferred by the AI exhibited the disciplines of foundational philosophy (philosophy of mathematics and philosophy of science), mathematics, epistemology, ethics, rationality, physics, communication, game theory, cognitive science, psychology/theory of mind, etc. that humans have.
It sounds interesting but I’m really quite far from having a concrete picture in my head—or even a vague outline—of how to write source code that would correspond to this text description.
Like, you say:
Perhaps the biggest problem is that humans themselves are not aligned on these disciplines…
Well OK, let’s just take one opinionated human, Alice, who thinks that they know what it means for a plan to exhibit the discipline of philosophy / math / physics / game theory / etc. or not. Let’s say she’s the only human who matters. We don’t care about anyone else’s opinions. OK, now we have solved the “humans themselves are not aligned on these disciplines” problem, right? But I still would have no idea what code to write, such that it would correspond to what you have in mind. Not even vaguely.
It sounds interesting but I’m really quite far from having a concrete picture in my head—or even a vague outline—of how to write source code that would correspond to this text description.
Like, you say:
Well OK, let’s just take one opinionated human, Alice, who thinks that they know what it means for a plan to exhibit the discipline of philosophy / math / physics / game theory / etc. or not. Let’s say she’s the only human who matters. We don’t care about anyone else’s opinions. OK, now we have solved the “humans themselves are not aligned on these disciplines” problem, right? But I still would have no idea what code to write, such that it would correspond to what you have in mind. Not even vaguely.
Replied here: “Aligning an H-JEPA agent via training on the outputs of an LLM-based “exemplary actor”.