Thanks!

I wonder if you wouldn't get better results by just grabbing the top system prompts for similar tasks and hard-sticking them onto the models, maybe with a bit of manual prompt engineering, such that e.g. Gemini always sees a very explicit order to assume user error when something goes wrong.

So this would forego our ability to assess how well they do autonomously and make the scenario more similar to having custom or dynamic prompting per task. That is also interesting, but aims a different objective, I think.

The other issue is that open-ended real-world tasks are a bit unfair to give to production LLMs, on the basis that anything these LLMs could do on their own would've already been done by an enterprising human using the very same LLMs, but with the intent to succeed[1] rather than the intent to evaluate the model's performance.

I'm not sure if just seeing how far they get is "unfair" but I agree they are much less likely to succeed indeed!

Thanks! Appreciate you sharing that

Thanks for sharing your thoughts!

I think even people who love themselves in a healthy way don't love all parts of themselves equally (ditto for how they relate to their partners. I love my desire for truth more than I love my neurotic insecurities about what parties I get invited to. I love my partner's love of learning more than I love his forgetfulness about house chores. I don't think that's wrong or unhealthy, and I think that there's massive gulf between getting problematically angry at my partner (or myself) for making a mess and acting like everything he does (or I do) is exactly equally lovable (or worse, thinking it).

I think this is a matter of definitions and differences in inner experience. The way I experience ‘love’ the way I described above there is no clear love more or less of a particular trait. I definitelt can experience more delight or desire or admiration or enjoyment for some traits than others but the way I use the word ‘love’ is more akin to a sense of connectedness and acceptance than how happy or appreciative I am about a trait.

As you note, not all of my genes show my phenotype. For example my partner and I could both have recessive genes that a child could inherit such that they end up with a phenotype that has traits totally different from ours. Some of those traits could be really big and really important (e.g. Tay Sachs)

Yeah, that part is hard… I think the more ‘costly’ the ‘hidden’ trait is the harder it is for it to be sort of ~overwhelmed by the remaining traits but you can still say you love all the remaining traits equally

Not all children are children of the same beloved partner. People have children by rape. They have children via one-night stands. They have children via partners that they used to love but no longer love. Maybe you'd bite that bullet and say, "Well I love all my kids equally, but probably parents don't love their kids equally in situations like that." (Which I'd find interesting, but seems like it's not the vibe of your post)

I think it is harder in those situation to come to love your kids equally cause you don’t have access to the above mechanic, but I 100% do think a lot of people achieve it anyway through other mechanics. I didn't mean to claim this is the only mechanic.

yeah, parents in general probably don't have much time to read. That's a good point!

I haven't tried most of the books people here recommend. I think I was mostly thinking of volume and how to find the books.

Thanks! <3 :D

Appreciate everyone's recommendations :) <3

I'm now also just trying to follow the age-old advice of well ... do it yourself then, so here is my attempt at writing something new into the genre that doesn't exist yet: Split (Part 1)

I neither act out my frustrations at other people nor represent the experience of time blindness that others may have. This essay was intended as an unfiltered inside look on a personal experience. Your comment strikes me as unnecessarily unkind.

I’m confused about your point… paraphrase? I don’t find myself having issues in meetings as people try to reach productive conclusions in meetings so you can optimize the process of the meeting around that. If a meeting would include people interrupting each other while reading or writing constantly then i would predict this to be a less productive meeting than one where people don’t do this as a matter of course. That does not seem specific to my experience of time blindness I think.

+1

And gosh… this is just what it was like growing up in a small rural town in the Netherlands. I was shocked when I moved to the city and people didn’t all greet me back and smile. I am guessing I am still comparatively trained to smile and greet strangers as if I am going to meet them regularly for the rest of my life and that seems like a good thing tbh.

I'd love to see the results of this :)