Crossposted from the EA Forum: https://forum.effectivealtruism.org/posts/DYmDGizxkcrZhmBCq/against-the-weirdness-heuristic

Slightly more than a decade ago (approximately 2009), I was imagining a pair of earphones that would look exactly like my iPod earphones but would have no wires. That pair was perfect in my imagination: it would fit in my ears just like my wired pair did, the sound would be great, and I wouldn't have to worry about disentangling wires. At the time, wireless meant "ugly Bluetooth (cool) kids never use". In 2016, Apple released the AirPods; they were pretty much what I had imagined plus you needed to charge them (I hadn't thought about that...)

About two decades ago, in the early 2000s, my father went abroad for a few months to work. We knew we wouldn't see his face for that period of time. I don't really remember calling with him either but maybe I'm confabulating here. His absence would have felt way less dramatic were we to video call. In 2010, FaceTime was released while Skype had already been around for many years.

I mention these two stories to argue that both "imagined" technologies became available very soon after I imagined them. I wasn't particularly a tech person, at least no more than the average child growing up at the time. I couldn't have predicted that these technologies should be available based on anything because, well, I didn't know anything. I was just thinking what it'd be nice to have and soon after this thing would be out for sale. (I personally didn't get AirPods till 2022, however).

Thinking about technology in the AI framework often creates the impression that the AGI future is weird. And if it's weird, it's also less likely to happen. This is the weirdness heuristic: I reject something not because I have evidence to do so but because we should generally not anticipate weird things to be true.

In retrospect, it's very easy to assume that having wireless earphones that you charge in a tiny case as well as video calling with someone on the other side of the planet whenever you want are both "normal", non-weird futures. But would we agree on that in the early 2000s?

Writing the "history of the future" would solve many of our problems. We would have clear timelines and plan accordingly. Maybe we would find that there isn't enough time to prepare or even if there is, that we still don't make it. Regardless, the arrow of time doesn't work that way.

All these thoughts lead me to this conclusion: don't resist having short timelines just because "it's too weird" (you may of course have other, legitimate reasons). If, as a kid, you imagined that the day you'd video call with your friends is somewhere in the far distant future and it happened within the same decade, you should expect similar "weird" things to happen in this decade.