It’s 2055, and you’re quantum-retired, a trendy term that a billionaire Zoomer coined, used to describe the millions of 60-80 year olds who both “work and don’t work”. (A few of the trendiest graybeards call it “cat retired” as a witty nod to Erwin Schrödinger’s famous thought experiment and/or the fact that we don’t really know what cats do all day.)
You’re telling stories to your family about Corporate America Back in the Day, reminiscing fondly on your early “career” as you bounce your granddaughter on your indestructible knee. She asks question after question about what it was like to always carry servers around with you — the legacy concepts of laptops / mobile / cloud / servers are fuzzy — and then the big question comes.
“What did you do at your job before generative AI?”1
You stop bouncing her and focus entirely on answering the question. You know it’s not as much about what you literally did at your job, like writing lots of emails and attending lots of meetings. It’s more about what changed. You sip your reishi-infused kava.
“We used to have to work really hard to be creative.”
Your granddaughter, a scary genius compared to the 11 year-old you, appears satisfied with your answer. These chats will really get interesting when she learns about Schrödinger in Classical Physics next year. She jumps off your knee to play with her robot bunny.
That is Scenario A.
There’s also a Scenario B, which is too depressing to fully write at this time, but the gist is that after you sip your drink (Jim Beam instead of kava) you reply:
“We used to be more creative.”
The recent explosion of generative AI makes for an extremely hard-to-predict future. While some people view it as just a better Google, others think it will lead to the end of all biological life on Earth.
Before running any sort of scenario modeling, it’s helpful to establish context. And the context here is the following: generative AI is the next tier-1, holy shit, worthy-of-having-a-dotted-vertical-line-on-a-100-year-time-series-graph technological thing. It is going to be more impactful than blockchain, 3D printing, cloud computing and maybe even the iPhone. Yes, “tier-1” as in, Internet and airplanes tier-1.2
When the Internet was the fledgling topic du jour in the 90s I was more into Pokemon than tech news, but the sense I get is that even though some people had a strong intuition for the potential of the Internet, almost every prediction of what the next decades would bring became laughably inaccurate. And this includes not just the talking head morning show muppets who got most of the airtime, but even technologists and their bubbly investors.
The point is not to laugh at them, because again, predicting the future is really hard.3 The point is that like the Internet, generative AI is a game changer in terms of how we will live our lives, which is fundamentally the sum total of how we spend our time. Consider how much the Internet changed the typical American adult’s day, in 1993:
…versus today:
We are basically 50% online creatures now. True, it took 30 years, but a crucial meta element of the Internet is that it not only spawns new technology, it also accelerates the adoption of new technology.
The Internet and Generative AI – same same but different
The tech adoption lifecycle is the classic framework we can use to contextualize a new technology’s ubiquity. ChatGPT, the most popular generative AI product, has 100 million users now, which sounds like – and is – a crap ton, but it’s still very much a “techie thing” at the moment. (Remember, there are 7 billion potential users.)
To give tangible examples, I’ve personally used ChatGPT in my day job to write emails and Slack messages, summarize full Zoom meeting transcripts, troubleshoot support in apps like Carta and Sage, and figure out some gnarly Google Sheets formulas. I don’t code (for now), but there are thousands of Tweets and Reddit posts about engineers using GPT to do their dirty work.
Generative AI ubiquity – which I will generously grant, is a couple years away – will further shake up the way we spend our time. Here’s a potential scenario that feels realistic to me (I’ve focused only on time spent at work, since that seems slightly easier to predict and more templatized than the rest of our lives):
But what exactly does this new mix mean? What is the impact?
The answer (which required some light gymnastics in the legend) comes down to a single word: creativity. The counterintuitive reality of having a new technology that ostensibly replaces humans and human creativity…is that it opens up room for more human interaction and creativity.
Let that sink in.
Generative AI is the latest in a long lineage of human-created technologies that ultimately, allow us to spend our time in better ways. The crazy thing is that it can do tasks that typically require human creativity, so in a way we’re raising the standards on what it means to be “creative”. That is amazing.
Of course, there are costs and risks. I strongly recommend reading Eliezer Yudkowsky’s plea to shut it all down and Tyler Cowen’s rebuttal, but the summary is that there’s a philosophical tension between “if we don’t get this superintelligence thing right the first time, there’s no do-over, because it will kill us” and “the future is radically uncertain, we cannot possibly orchestrate a centralized plan, and let’s dive into the future with confidence”.
The angle that does not get much attention and worries me slightly is that generative AI gets so good, that our relationship with it goes from symbiotic in nature (eg, have ChatGPT draft a statement, then the human curates) to completely overpowered (eg, some language model produces a song that tops the global charts).
There are two paths: either we up our game and keep up with the rising standards of creativity, or we become the humans in WALL-E, who endlessly consume content and calories. It’s obvious the former is better than the latter; as a rule of thumb, it’s better to create more than we consume.
Creativity is not just a feature of civilization – it is the backbone of it. And now we have better tools – and more time – to get better at it.
For brevity let’s assume “job” and “generative AI” are still used in the vernacular. I considered making up words and explaining them like sci fi writers do in the first few chapters of a story but 1) you didn’t come here for amateur sci fi and 2) I’m gonna retire at “quantum-retired”.
Here’s a delicious basketball analogy that you can ignore if you don’t like basketball: Generative AI is Victor Wembanyama, the French 7’5” 19 year-old alien that will be the first pick in the upcoming NBA draft. We have to go back 20 years – to LeBron James – to find a more compelling draft prospect. In this analogy, LeBron is the Internet, Anthony Davis is AWS and Zion Williamson is Litecoin (hehe). Where this analogy most prominently breaks down is that LeBron, unlike the Internet, was a sure thing. Let’s also throw in Lew Alcindor as the Turing machine.
But obviously you’re here reading this article, at least partly to laugh, so here’s a personal favorite: in a 2000 interview, our former President Bill Clinton, dismissing China’s ability to restrict free speech online: “Good luck. That’s sort of like trying to nail Jello to the wall.” A little dark, yes, but for whatever reason I get so much comedic utility from picturing him saying that, with the slyest of sly grins, emitting a confident yet man-of-the-people type chuckle between the sentences as he prepares the listener for the punchline, which he of course nails (woof), because he’s a case study in How to Be a Politician, and they just don’t make ‘em like that anymore, huh?