almusdives

The lost "work" is two years of ChatGPT logs. Sounds like AI systems had concrete benefits to this researcher in a number of applications, but I'm not sure how I feel that their discussions with AIs are so casually being described as "work". Seems slightly misleading?

Is there anyway to try this without logging in?

Funny how in a world of AI, human curation is becoming more important.

I appreciate that this figure is showing that the relative scarcity of gold compared to oil increasing, but I'm not sure what it's saying, or is interesting beyond that? Is this reflecting a global shift towards natural gas rather than oil, or reflecting investors are particularly nervous since covid, for example?

Obviously, you can't drawn any hard conclusions, but I was wondering what the OP was thinking narratively when posting.

Definitely says something, but I'm not sure what. How are you interpreting this?

Just wanted to say this post has caused a huge spike in traffic to my language's website: a dizzying ~40 visitors per day up from ~0 haha!

As a scientific researcher, I use LLMs all the time. Mainly in place of Google search, to help write code and maybe summarize a paper here and there etc. But I definitely don't use it for the actual scientific process e.g. hypothesis generation or planning analyses etc. It tends to produce a lot of vague bullshit for this kind of thing; while not wrong not entirely useful either. I have a few colleagues that do though with more success. Although I think the success comes from articulating their problem in detail (by actually writing it out to the LLM) which I think is the source of "inspiration" rather than the resulting content from the LLM.

I get why this is problematic for industries that depend on high traffic for ad revenue etc, but is bad for websites who are actually trying to market services that provide tangible value? Like if I’m searching for a dry cleaner in Glasgow, if I end up with the same provider, I don't care (and neither does the dry cleaner) whether I find them through traditional links or an AI-mediated search?

Over the last few years, I’ve been working on Zyme, an esoteric language for genetic programming: creating computer programs by means of natural selection. I’ve started seeing promising results, showing that random bytecode mutations can, over time, lead to measurable improvements in program performance. While still a long way from state-of-the-art approaches like neural networks, I wanted to share my progress in a blog post.

Feedback and criticism are welcome!

> We actually know exactly which gene sequence is responsible for real intelligence.

We don't at all know this.

You're right - a disassembler would be incredibly valuable; and while I haven’t gotten around to implementing one yet, it’s definitely on my radar. But even the idea of even having an assembler is exciting because it suggests the output programs could be interpretable, enabling us to identify the underlying algorithms that (hopefully) emerge through evolution.

Thanks! Since the language is still in its early stages of being able to evolve programs, I don't have many examples to share yet. I didn’t realize when I began but developing the language itself was just the beginning - I hadn't anticipated how much work would still be needed on tuning, development tools, and implementing the genetic programming framework before getting concrete results.

Sadly not. When I was developing Zyme, I was thinking a lot about the molecular components of a cell and how one might translate them into a virtual machine. I was particularly inspired by enZYMEs.