Does Lesswrong allow GIFs?
With the recent emphasis on effectively explaining AGI to people who don’t already know about AGI (which should have happened decades ago), I think that imagery is something seriously worth thinking about.
I should be able to cleanly place this GIF in between my paragraphs. It has zero communicative value as an image, and it doesn’t prove my point efficiently if I leave it as a link. Did you click it? Or did you feel slightly befuddled by the green text that could communicate all sorts of things but failed to do it here? If you’re anything like me (which you shouldn’t aspire to be), then you probably did the latter.
If you’ve clicked it by now, you probably were pretty glad that you did; it’s a pretty fantastic gif after all, and you can send it to your friends and family and put it in reports and all sorts of wonderful things to reduce the risk that the people in your life lean on the Absurdity Heuristic to evaluate their relationship with you.
But that’s not the point. You shouldn’t have needed to click the link to that gif. That’s not how reading is supposed to work. Humans didn’t even evolve to read, writing just took a form that works with our ape brains. And even if humans did evolve to read, the mutations probably wouldn’t have resulted in anything like “optimal reading” anyway. Communication has always had to adapt to suit the meaty computer, whose basic architecture is mostly static.
There are plenty of other circulation-worthy gifs that are even worthy of circulation on Lesswrong (and yes, I know a lot about the pitfalls of image circulation, including plenty of infohazards that I don’t dare to mention on a public forum post). An example I can think of off the top of my head includes this one. If Lesswrong guaranteed that it would be worth my while, I could potentially even make universe-saving gifs out of videos like this video.
Gifs might not be compatible with reports or policy briefs, but they can certainly go a long way towards explaining bits and pieces of alignment to people without them freaking out about the person telling them that a supercomputer will one day spawn an unstoppable eldritch abomination, which proceeds to ruin everything for everyone forever, and the only solution is to give some people in SF a ton of money. This isn’t about mass public outreach, which is horrible and should never happen due to second-order consequences beyond anyone’s wildest dreams, but gifs do fit great into emails, and emails fit great into the need to sensibly explain sensible concepts to sensible people.
I’m very sick and tired of living in a world where there isn’t an email template that can successfully explain the situation to someone >40% of the time. We should seriously consider the possibility that we’re collectively overestimating the difficulty of explaining AI alignment itself.