I have a long list of topics I'm planning to learn more about in my free time and I'm curious what kinds of things HN readers would focus on.
This is already one of my favourite things in life.
Projects and companies come and go .But the focus value from this I find invaluable.
AI. I think we're getting to the point where we've seen just enough progress with contemporary AI systems, to think that the "knee of the curve" is close, and that within a few years "everything is going to change." I'm at a point with this to where I mostly think that working on anything other than AI is probably a waste of time.
Of course in my case AI has been one of my top interests for the last 30 years anyway, so this isn't some big swerve for me. But it motivates me to "lean in" to AI research as fully as I can, and to focus on AI applications, integration, etc.
OK, so saying "working on anything other than AI is probably a waste of time" may be a little bit of hyperbole, but I definitely see it being pretty damn important. That said, if I was going to carve out some time to study anything other than AI, for me it would probably be biochemistry, bioengineering, nanotechnology, or electronics / hardware stuff.
In the last several months I have used ChatGPT for things like thought experiments and churning out code faster.
That's tough to answer. I have some specific areas of interest that I've cultivated over years and decades of maintaining a somewhat passive interest in AI, and only starting to really work on more serious research over the past few years. And even then, my list of "stuff to look into" is somewhat speculative, so I'm not sure how much value there would be in going into it in detail.
That said, I've alluded to this briefly in previous HN posts, and it's not any kind of secret or anything. My interest areas include a heavy-dose of focus on "neuro-symbolic" systems: that is, integrating neural networks and "symbolic computing" (aka "GOFAI" or "Good Old Fashioned AI). To drill down a bit more: the idea is to use neural networks where they are really good at pattern recognition: image recognition, speech recognition, some language understanding tasks, etc., and then integrate that with more purpose built systems that work at a "symbolic level" (or you might say "knowledge level") to do things like deductive reasoning, abductive inference, temporal reasoning, inductive inference, common sense reasoning, etc. One of the tripping points for this is how to translate between the representations of "knowledge" across those different modalities.
In the last several months I have used ChatGPT for things like thought experiments and churning out code faster.
Yeah, I think for anybody who maybe isn't interested in diving into AI research per-se, doing stuff like that is very valuable. At the very least, learning to use and manage and work with the AI tools/platforms that are put out there will probably be an increasingly valuable skill over the years to come.
The only thing I'd add to this is to say that I think it might be a mistake to focus only on LLM's. Don't get me wrong: I'm amazed at what LLM's can do and I think they are - and will continue to be - very important, but I don't know that I think they are the "be all, end all" of AI. It might be worthwhile to look at the suite of AI/ML offerings being provided by cloud providers like AWS, Azure, GCP, etc. and at least get familiar with working with some of those things.
One domain I've encountered that falls into this category is mom and pop food stores which don't typically have a decent website. You typically have to visit the store or call to find out things like what they stock and current prices (to do comparison shopping).
Still, you asked what I did. I read, a lot, and deeply. I wrote tens of thousands of words about what I read and did. I guess you could say I went into myself.
I did manage to do a little road tripping, carefully, avoiding crowded tourist venues and mostly seeing the country.
I did a lot of photography, my hobby-job.
Thanks for the suggestions. I do have a pretty decent woodworking shop and plan on spending more time there.
Right now GPT and Stable Diffusion are very popular and there are many potential new businesses.
My suggestion is to be very careful which GPT you use though. There are a lot of inferior ones out there.
Very grounding. Still requires mental acuity but gives you the opportunity to use different parts of your brain, and it rounds you out as a person.
Very grounding and it has been good for my mental (and physical!) health too.
take the summer and chill, without any guidelines or expectations. this may feel weird at first. creativity is the residue of time wasted.
then do some more structured and ambitious adventuring in the fall. if that feels annoying, continue to chill.
in winter, start thinking about which kind of people you’re interested in working for, and try to figure out where they are and how to reach them.
tldr; summer is just the beginning!
wrt to technical topics to mess around with, follow your own curiosities. for me it’s gamedev, ops/infra, and data/secrets management.
it looks like you’re interested in rust. in case you haven’t seen it, checkout #showcase in bevy’s discord server. lots of people posting cool stuff in there.