There’s so many interesting things happening on the science/science fiction beat these days, it’s hard to keep up with. Here’s a few of the choicest cuts I’ve digested in the last few weeks. Enjoy!
The Creator is coming out next weekend, and I’m feeling pretty geeked. It’s written and directed by Gareth Edwards, who also wrote and directed the best Star Wars movie ever made. Expect my review in two weeks in this space, and please go support the movie in the meantime. Original sci fi films by great filmmakers shown in actual movie theaters are tragically rare these days. I can’t wait.
DALL-E-3 is going to be available in early October. Here’s the Youtube trailer for those who prefer to watch. Here’s my post on DALL-E-2 for those who need a refresher. Expect a full post from me the Sunday after it actually drops. Two rumored improvements I find very exciting: A) The ability to iterate a piece through multiple prompting rounds (e.g. “Make it just like the last one but make the sky orange”). B) It will be integrated directly into ChatGPT+, meaning it will have a profoundly powerful language model at the front end. This is the beginning of the consolidation, the endpoint of which is to have one app that you just talk to in plain English and it can do for you anything that’s possible to do with a computer.
Speaking of which, GPT-4, access to which is now available as ChatGPT+ for $20 a month, is something that anyone who does knowledge work of any kind should pay for access to and be working regularly to use to improve their productivity. We are right at the inflection point, where in the next ~5-8 years A.I. is going to go from “cool but not obviously game-changing” to “THE tool that separates the indispensable from the obsolete in competitive industries”. It’s before you really need to know it that the time is ripe to learn it. Yes, it’s non-obvious how to start, but that’s what makes doing it so valuable.
Toby Ord is one of the founding members of the Effective Altruism and Longtermism movements. He’s an ethical philosophy professor, and an expert on existential risk. What I didn’t know is that he’s also curated and retouched the world’s most extensive collection of photographs of the Earth taken from space. He’s got detailed information on each one, including many I’d never seen before, and a really slick web site displaying them. Just a pleasure to explore if you’re into that sort of thing.
This episode of the 80,000 Hours podcast isn’t new, but it was new to me this week, and provides the most convincing answer I’ve ever heard to a question that has long mystified me: “Why is it so hard to do economic development in poor countries even with massive amounts of money available to spend on the problem?” It’s not science fiction, but it’s well worth your time even at a capacious runtime of over three hours.
Hall-of-Fame Youtuber Kurzgesagt in a recent video answers another question I’ve idly wondered about for years, in his usual inimitable style: “Can you destroy a black hole and what would happen if you did?”
Deepmind has a new feature that is supposed to be able to help identify A.I.-generated images, watermark them, etc. Color me skeptical at most margins. Will large companies use this to safeguard their IP? Undoubtedly. Will this prevent misinformation or abuse of generative A.I. by highly motivated bad actors? If history is any guide, then resoundingly not.
Hope you found something educational or informative in those! Have a great week, and I’ll be back next Sunday with another original story for your enjoyment.