“We wanted flying cars, instead we got 140 characters” — Peter Thiel
“I’m so happy I lived long enough to see this!” — My dad, watching his 6 year old grandson use ChatGPT
Bit of a change of pace, but not really. I think a lot of this site explores the ongoing story of social media’s impact on the sports world. Speaking beyond and including the sports sphere, I don’t think we were designed to handle this level of interconnectedness. Its advent has led to the sorts of destructive media hysterias chronicled in Nellie Bowles’ new book. Unlike, say, Tyler Cowen, I do believe that social media technology has been a net negative so far, even if it’s helped my career. It’s just how I see it, after witnessing so many otherwise intelligent professionals lose their minds. Movies in the 1990s were superior. The Before Times were better, in many ways.
But I am not a tech doomer. And I am not an AI doomer. I am indeed wildly optimistic about AI and less focused on the potentially devastating downsides. Why? First, there’s the secular Pascal's wager view on this. If AI really is a powerful enough force to kill us all, then such a force seems rather inexorable. I certainly don’t trust some group of modern humans to keep that potent a pandora’s box sealed while reaping the benefits of what’s inside. And people are going to chase those tremendous benefits, so there’s no sealing the box. No matter what constraining rules are proposed, someone will build on today’s advancements in Artificial General Intelligence. This is a collective action problem if you see it as a problem.
So, going with The Wager, I just choose to believe that the tech apocalypse probably won’t happen. Instead, I choose to believe that this technology will unleash discoveries that benefit mankind. I’d say the latter is already happening, at least. You can tell me I’m wrong and I won’t fight you much. But here’s what I’ll say about how amazed I am by the inclusion of this technology in my house…
I downloaded Chat GPT to mobile and it’s the first invention to inspire true wonder across all generations of my family. That voice feature might be creepily reminiscent of the movie Her, but my 6 year old has no such reference point. He’s just thrilled that there’s now a teacher in the house who can instantly answer all his queries.
My boy is prone to blurting rapid fire follow up questions at a high volume, which is endearing to me, but perhaps sometimes suboptimal for a school setting. I’m generally not into revealing details about my family, but some of you have kids, and might have encountered similar challenges. School has its benefits, but it’s more an uncomfortable energy restrainer than an energy harnesser. It’s important to learn how to sit still, be quiet and take in a lesson, but that’s a tougher ask for some kids. For us, GPT has offered an educational outlet, away from the grind.
With GPT, there’s no reason for him to sit still and contain himself for the sake of everyone else. He can just indulge his excitable curiosity to the fullest extent.
I tell GPT to pitch explanations to a six year old level and it’s a glorious reminder of Richard Feynman’s, “If you can't explain something in simple terms you don't understand it.” It actually works! Seemingly abstract complexity shrinks into tidy metaphor. The Machine plainly conveys why an atom can’t split when you cut your hair, how the tapetum lucidum allows our cats to see in the dark, and what makes that loud sound when you pop a balloon. And then if we want GPT to make the explanation into a rhyming limerick, that’s no problem either.
I’d be lying if I said the main use of AI in our house is educational. Instead, the Machine is often asked to create new versions of Pokémon themed stories. Characters get woven in as my child offers pointers on motives and scenarios. I missed the Pokémon craze as a kid but thankfully Chat GPT has access to the vast backlog. Sometimes I’ll direct the device to involve my son in the tale, which he seems to enjoy. When we were driving to the farmer’s market, with my father, I made such a suggestion. Pikachu met my son in a story that featured many landmarks familiar to all of us. The tale was told so well, so seamlessly, that it inspired my octogenarian father to exclaim:
This is amazing! I’m so happy I lived long enough to see this!
I’d never heard my father say that. Actually, I’d never heard anybody say that, at least in person. We spent much of the day letting The Machine lead our inspiration. Unlike playing a video game for hours on end, or scrolling Instagram, the experience didn’t feel dirty. None of that afternoon felt like a waste of time.
This is the part where you might bring up concerns regarding this burgeoning technology. There are tradeoffs. What I’m describing certainly seems like a potential threat to teachers. Other sorts of jobs are and will be threatened. I don’t want to be glib about these possible outcomes, especially after watching GPT take a crack at trying to do my job in my style. I also can’t predict what industries the tech might give rise to and whether more jobs are created by whatever we learn.
It’s just hard for me to see this development as overall bad. Knowledge is good. Improved problem solving is good. If cancer gets cured, it’s probably via AGI tech.
What I do know is that the era of, “We wanted flying cars, instead we got 140 characters” is over. That Thiel quip aptly summed up popular disenchantment with this “tech” of the modern age. Where’s the splendor? Where’s the awe? Why are all the major companies devoted to crassly manipulating peoples’ attention spans?
No, that era done, even if the biggest invention casts no vast shadow in the physical world. As I drove with my spellbound father and son, doing a suburban 20 MPH, floating through the world of Pikachu and his friends, the car was indeed flying. Finally, we could all travel wherever we wanted, be it the universe of Pokémon, or actual outer space. Finally, we could all explore unbound. Finally, my son could take the wheel.
Not a doomer, but yeah…no. It encourages further passivity and trains people to be receivers, this is an especially subtle problem for young people (broadly defined) who are in the prime of development.
Acts like creating a research plan to get an answer to a question and then having to research an issue by cross referencing sources and then synthesizing them (using your judgement throughout) or creating your own story that incorporates your life into the fantastical to amuse yourself or in general learning how to use your imagination when bored, create immense value that we underestimate.
It’s great that we can harness AI’s power for greater efficiency in professional and research settings, offloading certain necessary but rote tasks to it…there are always pluses and minuses, but I have doubts we will use this technology in a way where the former will outweigh the later.
I’d be very careful with how much I let my kids use Chat GPT to learn things because it’s wrong.. like a lot. “Hallucinations” are way more common than people think and tools like Chat GPT aren’t designed to be “right” they’re just designed to sound right.
Also, I just read a fascinating series on AI in Ed Zitron’s newsletter that highlights we may actually have already seen peak AI because these large language models have already consumed basically the entire internet and would need 5X more data to continue to evolve. Pair that with the limits on computing power being nearly reached and I’m not nearly as concerned as I used to be about AI taking over the world, I’m more concerned now that AI is just the new crypto and the bubble is gonna burst.