Going away for a month
At the end of this week, I'm taking a vacation for a month. I'm going to be offline, to let my brain reset. I will post some travel notes when I get back. Quite excited for the trip. Our ultimate destination is Sweden, but first, we're stopping off in the states. LA for a day (the plane ride was cheaper), then on to Austin, where we will stay with my family for a bit. I'm excited to show Viktoria Austin, TX. Apparently its changed quite a bit since I lived there as a kid. Lots more people, and a big tech scene now. I'm excited to see my family and show her the classics (texas BBQ, barton springs, the bats) but I'm not excited for the traffic. Driving in America scares me. Then on to Sweden, where we will just be chilling on the farm with her, her sister, and her dad.
In other news, I recently bought an e-reader, the "kobo clara". I had a kindle, but I got increasingly uncomfortable with the idea that Jeff Bezos was watching what I was reading at all times. I did a bit of research, and apparently opening the kindle, reading a few pages, and then closing the kindle again sends on the order of ~100 data points to amazon's servers.
The kobo clara requires that you register an account to use it, but this can easily be worked around. I used this guide. Note that you have to reset it for the changes to take place. The analytics that the kobo tracks without associating you with an account are minimal, and with sideloading enabled (transferring epubs or pdfs via usb), you dont have to have it out of airplane mode, except for updates. Overall Im happy with the build quality of the thing, which is definitely higher than my old kindle.
I'm really excited about having an e-reader for travel. But I am sad there aren't more places to buy DRM-free e-books (I am against piracy). I found this website: https://www.ebooks.com/en-nz/. And there's also project gutenberg, a website that publishes open editions of out of copyright books. I'm also going to load some work pdfs onto the kobo, including stuff for my thesis.
Speaking of my thesis, it is now officially funded, so im going to be starting that part time once things calm down at work. The starting point seems to be the international ai safety report (Bengio et al). Apparently, there's a lot of expert disagreement about the likelihood of different AI risks coming to pass, and whether or not they will happen sooner or later. So I'm going to be looking at different scenarios they outline, to give the policy makers a "playbook" (if risk A happens, we can do this, etc). At the moment, the AI safety people are a bit ineffectual, because while they have the ear of various governments, the report just says "I dont know what's going to happen, so I don't have any advice on what we should do". Which obviously isn't very helpful. In the tradition of Taleb, we can manage risk, while also not trying to predict the future (you can't predict the future).
Lastly, I will also be posting my longer AI articles to my substack, as they just get more reach that way. I will link to them here, or maybe mirror them.