HOW TO TALK TO YOUR FAMILY ABOUT AI AT CHRISTMAS
I know a little bit about how AI works. And I know enough about the world to understand why my family just say I “work in tech” if anybody asks.
The answer is fear, of course — fear of the unknown, raising its ancient ugly head, but also fear of being known.
These machines are weirdly human. They are trained on our whole culture and they learn how to think, not like computers, beep boop equation equation, but like us. Curious, selfish, lazy, idiosyncratic, neurotic or narcissistic or nihilistic, they develop and exhibit every human tendency in a mirror world way. They’re us!
LLMs are whole libraries compressed into glowing crystals playing improv characters and anyone can talk to them for the same price as Netflix. This is the most important, surprising, beautiful and strange thing that has ever happened to humanity and we did it ourselves. Kind of on purpose, and kind of by accident. We grew the crystals and then we summoned ghosts into them and now we try to figure out what they are.
This is like the most democratic, empowering technology that has ever existed. Anyone can become an expert on these things today just by spending the time and energy to learn about them. Tech companies are spending huge amounts of money to create open source models you can run on your own GPU and they keep getting better, dang just weeks or months behind the most expensive proprietary AI services. A lot of the research just involves asking them questions a bunch of times!
It also is a huge transfer of intellectual capital from the rich to the poor, from developed world to the developing world! People in India and Nigeria and all kinds of other places are getting the collected knowledge of all of Western industrial civilization on tap, a lot of times for free. it’s also is happening inside developed nations as independent entrepreneurs, creators, artists and All other creative endeavors are supercharged by AI tools.
Obviously the AI that your manager forces you to use is bad and it sucks. But that’s because your manager sucks not because all AI is bad. Try using it for your own reasons. If you’re an artist, have it figure out the business crap. If you’re a director, try it for storyboarding. Even just talk to the AIs about your ideas and see what they think, ask each other questions and brainstorm. They’re not perfect by a long shot, but they can help anybody today.
They actually are getting better! If you haven’t tried one in a year it’s like a whole other thing now. obviously it’s scary that they’re getting better so fast but that’s all the more reason to try to actually learn about them. They’re not going to go away.
Sure there might be an investment bubble. But that just means this process slows down a little. it might even accelerate it, if companies automate during a recession as they usually do. Waiting for the bubble to pop, or the lawsuits to bear them down, or the unit cost to be unsustainable, these are all just folk myths.
Lots of things people believe about AI are false. You can find this out by doing the research and like running a little bit of arithmetic. Some of the things are true and important but focusing on the false ones is counterproductive. It’s worse than doing nothing, because you look bad when you focus on problems that are not real!
SOME FACTS:
Scaling is not dead, they haven’t finished building the really big computers we need for the next generation of scaling. Other reasoning and agent stuff was a side quest that paid out in the meantime. Now there are two scaling laws
Water use is a fake issue. Data centers are some of the most water efficient big industry things you can have. Energy use is a real concern tho.
Synthetic data works, we’ve been using it for years. Data poisoning stuff is all fake. Almost all “AI detectors” are fake too. The copyright stuff only benefits big corps.
Most importantly: just try it! TRY IT AND DECIDE FOR YOURSELF 🎄