I don't really do "New Years resolutions", but I do (sometimes) do at least a little high-level planning for the year ahead. Coincidentally, I just did that exercise a few minutes before seeing this post. Well... some of the "planning" had been going on in my head for a while of course, but I mean I took the time to actually write it down.
I'd just share the doc with y'all but too much of what is in there would be really opaque because it refers to labels that I put on things that would be meaningless to anybody else. But as a summary, I have sections for
Reading / Research -- refers specifically to books/papers I want to read and research I'm working on.
Learning -- refers to "basic skills" or what-not that I want to focus on.
Building - refers to specific projects I want to work on that involve building something (could be hardware or software, or a combination of both)
Fogbeam - refers to anything related to Fogbeam Labs, my company.
The first section, Reading has reference to a number of books on linguistics / language / computational linguistics / NLP that I plan to work through. It also mentions reading the remaining books by Ben Goertzel that I haven't already read yet. Same for Ron Sun. There's also a section on Machine Learning and Evolutionary computing (Genetic Algorithms, Genetic Programming, etc.). And then there's some miscellaneous stuff. But realistically, I won't get through all of it in 1 year. I finished 50 books in 2021 (which is around my norm) but that includes quite a few novels and shorter books. Most of these are denser / meatier material that will be a slog to grind through. So I'll probably be lucky to finish 30 books in 2022.
Anyway... in the Learning section is some stuff about a sequence of math topics I want to focus on. I won't bore you with the details, but just suffice it to say I feel like I need to brush up on some forgotten math stuff and learn some stuff I never studied in the first place.
There's also a note here to continue working on learning more about working with the SOAR cognitive architecture.
In the Building section is a mention of an (e)-book manager / paper manager / note-taking app that I've started building for myself. Yes, I'm building my own because I have some very specific ideas about how I want it work, to suit the way I like to do things. If I ever get a reasonably decent version finished I might open source it. Most of what I work on is OSS, and that tends to be my default mode. But we'll see.
The other thing in this section is a "body" for some of my AI work. By "body" I don't mean a biped, walking/talking, anthropomorphic body... no, I just mean an old boom-box that I gutted out, and into which I plan to install a raspberry pi, a battery pack, a pan/tilt controller with a web-cam, a microphone, and some other miscellaneous sensors (accelerometer, temperature sensor, etc). It's meant to be "just barely enough" of a form of embodiment to let me do some experiments with how an AI might learn from interacting with its environment. Here is where I should mention that my reading list also needs to include a bunch of stuff I plan to read about developmental psychology, first language acquisition, etc. The idea is to play around with ideas related to the whole "let the AI learn (sort of) like a baby does" meme.
And if you're wondering, no, the rPi isn't meant to do the "heavy lifting" in regards to the AI "stuff" associated with the box thing. It's just going to be a front-end, collecting signals from the various sensors, possibly doing some pre-processing, etc. and then forwarding everything to another box where the real work will be done.
Under the Fogbeam label is a bunch of "finish product X", "finish product Y", type stuff. I'll say no more about that for now. But one of the bullets there is to start blogging more again, so hopefully I can make that one stick and maybe some posts from the Fogbeam Blog will start showing up here on HN in 2022.
Happy New Year everyone, and best of luck to everybody in the year ahead, regardless of what your plans are. (Unless you're planning something evil. Boo. Don't do that).