> The models do not have a fear of future regret.

I so much feel this specific point. All models up to now (including astra, fable) are too much trained to "get the job done" and pass the benchmark that its doesn't care at all on what happens after.

I'm just wondering why no one tried to RL a model on stuff like "less LOC" and "less overengineering", "use what is available in the environment instead of reinventing the wheel", "don't look for dumb corner cases" ecc.

Existing models can be steered, to some degree, but it's a continuos fight. Even if with specific skills/prompts.

▲

sebzim4500 an hour ago | parent | next [-]

I'm sure many have tried but it sounds pretty hard. E.g. optimising for less LOC will lead to horrific code golfing. In reality you need a very complicated optimisation objective that trades off all those factors, I'm not surprised it hasn't been solved.

	▲	vb-8448 an hour ago \| parent [-]
		My experience up to now is that the models can do both "less LOC" and "clean code" at the same time, you just have to keep reminding it to them. So the capabilities are definitively there.

▲

conmod278 an hour ago | parent | prev [-]

Eliezer Yudkowsky – AI Alignment: Why It's Hard, and Where to Start

https://www.youtube.com/watch?v=EUjc1WuyPT8

Eliezer talked about these ideas way before everyone else.