I wrapped the first version of my new book Safe For Humans AI A "humans-first" approach to designing and building AI systems (free to read online link) last Friday so I am taking some time planning on what I want to work on next. I don’t count consulting in my planning since I no longer accept development work and now mostly only act as an advisor. I now spend less than ten hours a week doing paid-for work.
I have a slow-changing work setup, but I may have to make changes to get more personal GPU compute. I am looking at both hardware purchase and cloud compute options. I have been relying on a Colab Plus account and renting Lambda Labs GPU servers for an hour or two hacking sessions. Replacing my 4 year old home GPU rig (which I seldom use because it is so outdated) would cost at least $5K so sticking with cloud compute would be much cheaper, but less fun.
Technology aside, I think the most important factor in having an effective work environment is time management. I try to have a few specific intentions each day, things that I want to accomplish. These intentions may be things like researching something on a TODO list, a writing goal for the day, a planned hike with friends, a large block of time reading, work on open source, and having meals with family and friends. My intentions vary a lot from day to day. I find it easier to be productive when I plan daily work activities on an equal footing as my “living as a human” activities. My daily goal is simple: to complete most if not all of my intentions for the day and to enjoy every activity equally. I have two warning signs that I pay attention to as signals I need to correct something: if I find myself not enjoying what I am doing in the moment or if I am bored.
Never OWN rapidly depreciating assets.