A few years ago, we had two new disruptive technologies hit the zeitgeist level of conversation: blockchain and AI. I was already tracking the early stages of both, but when the early versions of ChatGPT were released, it was immediately clear to me that blockchain would fall to the background and AI would soon become the conversation.
Why? Because I saw that AI was going to be immediately useful to everyone, whereas blockchain felt (and continues to feel) like a niche solution looking for a mass problem to solve.
I also immediately knew that some components of my author and advisor jobs were in jeopardy and that some components of those jobs would become amplified and a lot easier to do. Even harder for me, I saw that my teammates would be even more imperiled, as AI would be able to assist me in many of the ways they were assisting me.
I figured I had at most three years to adapt or become obsolete. I immersed myself in how I was going to use it and how I could help my clients use it. I’m glad I went on that deep dive because I overestimated how long we had.
Along the way, I read the intriguing “A Coder Considers the Waning Days of the Craft,” which introduced me to the chess world’s use of the word centaur.
In chess, which for decades now has been dominated by A.I., a player’s only hope is pairing up with a bot. Such half-human, half-A.I. teams, known as centaurs, might still be able to beat the best humans and the best A.I. engines working alone. Programming has not yet gone the way of chess. But the centaurs have arrived. GPT-4 on its own is, for the moment, a worse programmer than I am. Ben is much worse. But Ben plus GPT-4 is a dangerous thing.
The centaurs have arrived in every field and will continue multiplying because it’s a near inevitability. If a centaur will be able to beat the best humans, the only way to be competitive will be to become a centaur.
It doesn’t matter what your profession or job is. A centaur will eventually be better at what you do, because the best non-centaur humans will become centaurs and the average non-centaur humans will become centaurs, too.
When I read that article, I saw the inevitability for me as an author and advisor: I, too, was going to need to become a centaur. Fast.
So that’s what I’ve done.
If I were writing this article last year, I would’ve said I’ve got good news and bad news about the transformation. The bad news would’ve been that you’re in the same boat and you will have to adapt and transform. The good news is that it may actually get you closer to your best work, faster and cheaper.
Both are still true, but the only real difference is that the bad news is no longer really news. That’s how fast things are moving in this space.
I have no desire to join the hordes of AI productivity bros or prompt kings. They’re probably already in your feeds precisely because they’re using AI to 10x their ability to be everywhere, all the time. I won’t get into specific tools because anything I say about them will be obsolete in three weeks.
Instead, I want to provide a subtle shift in the way you may be thinking about AI that’s been really helpful for my clients. Instead of thinking about the ‘a’ in ‘AI’ as ‘artificial’, I suggest thinking about it as ‘assistive’ so that you remember that its job is to assist your core best work.
Thinking about AI this way brings home two important points:
You are not trying to replace your core best work. Whatever infinite game you want to play, AI can lessen the parts of that game that you don’t want to do or don’t want to learn to be good at.
Instead of thinking about all the ways you could/should/might be using AI, you can reframe the question to: “What assistance do I need to do more of my core work?”
It may not seem that mentally substituting ‘assistive’ for ‘artificial’ would make that much difference, but it’s been a mental construct that’s helped shift clients and colleagues from feeling AI-overwhelmed to becoming a centaur overnight. Whenever you see the next AI tool, I want you to be able to quickly filter if it’s useful to you — by determining how it can assist you rather than whatever random thing it can do that may not be relevant for you. Once you get AI to help you do your core best work, you’ll have more time to decide whether you want to learn more skills or just get better at your core best work.
I’ll give a brief example here, using writing as an example because it’s the lingua franca for so many of us.
If a blank page is an obstacle to you getting writing done, your goal is going to be to use AI (remember, assistive intelligence) to help you get some words on the page. In this case, all the fears about how poor the writing quality of AI actually misses the point: you weren’t trying to hit a few buttons and get yourself out of writing, but to hit a few buttons to get yourself to write.
On the flip side, let’s say that editing your work is what keeps you stuck. In this case, you would be using AI to edit your work. You could create bots that do copyediting or line edits, or you could build bots that do more structural editing. You could have one bot be both or two bots that you work with in different stages of the writing.
We’re each going to need to create different bots to assist us in different ways because we’re all unique. As someone who loves building systems of work, I find this to be the most thrilling part of the centaur life.
For instance, AI has been really helpful for me as a research assistant. From about 2009 through 2018, I didn’t have a good system for saving what I was reading, and to be honest, it’s still not that great. But I can usually remember when I read something and what it was, so I can ask my research bot to find sources in that time period that may be what I read. It’s right 90% of the time, which is better than my not being able to remember or participating in the oh-so-common online creator pattern of just passing off someone else’s idea as my own.
The main conversation about AI is focused on work, which makes sense because work equals money for all parties involved.
However, AI is even more useful to me for personal assistance and everyday tasks.
Here are some of the ways I’ve used AI as a personal assistant and to assist with everyday tasks:
I’ve built a bot that can help me determine whether taking our Telluride, our motorcycle, or renting an RV is going to be the most cost-effective way to do any given trip. It can’t tell me which is going to be the most enjoyable, but it can do a good job of displaying the costs and timelines based upon factors I’ve fed to it.
When we were doing the National Park pilgrimage this year after Petie died, Angela used a bot to help find places to stop while she was navigating and I was driving.
I have a bot that can help me plan what I’m batch cooking on the grill.
I have a bot that helps me find ETFs (Exchange-Traded Funds) based upon sectors I’m interested in and can do a lot of what used to take me a long time to do on spreadsheets.
I use ChatGPT to help me shop. Instead of searching for products, I can tell ChatGPT what problem I’m trying to solve and give it some parameters, then it will search for highly-rated products that solve that problem, as well as the pros and cons of the product.
In general, ChatGPT has replaced the Google Search bar for me. The more I use it, the more it “knows” about me and can provide better answers.
Someone reading this is yelling to themselves, “What about AI hallucinations?! What about its repetitiveness and inability to do math consistently? What about the bias built into the models?”
Generally, the more high-stakes and important something is, the more I either won’t use AI or double-check the results. But that’s really no different than how I interact with humans because humans misremember, make stuff up, and can’t do math consistently.
Additionally, I have no expectations that any assistance AI gives me will be perfect or a replacement for my own judgment. I know I’ll have to give it many more prompts to get to something I can use.
But I’m such a contrastive thinker that it’s great for me to see something, be able to react to it, and ask a better question or give better instructions. Instead of something just rolling around in my head for months or years, I can use AI to help me clarify, distill, expand, or decide about something.
Lastly, when it comes to AI bias, let’s remember that AI was trained on our biases. AI is about as biased as humans and I’ve added parameters to my prompts to account for this. The AI bots I’ve trained consistently check their biases. It’s so much harder to get humans to do that.
The real challenge that most people aren’t talking about yet is skill decay. After spending several months experimenting with AI in my writing flows, I noticed some skill decay in creating drafts without AI and the depth and voice of my writing were suffering. Given how critical writing is to me personally and professionally, I knew I needed to course-correct quickly.
#SorryNotSorry that you’re stuck with my idiosyncratic longer posts that violate so many best practices of online writing and instead are getting more ensouled and ‘unnecessary’ words. (I’ll likely write more about how to avoid skill decay since it’s the danger I’m most concerned about for all of us.)
Like any technology, AI has limitations and dangers that we should always be mindful of. But to my point above, it’s inevitable that most of us are going to have to become centaurs to thrive at work.
What most excites me about AI, though, is that its real promise is assisting us to be better humans.
Parents can use AI to help them help their kids with homework.
Partners can use AI to help them communicate with their partners.
Leaders can use AI to help them lead and manage better, which will make work better. (More coming on this in a future post.)
AI can assist people with their hobbies and creativity and unlock new parts of themselves.
AI can help people diagnose their difficult medical matrix, help people understand their diagnoses, and/or help doctors explain diagnoses better.
AI can help people understand their spiritual texts and traditions.
Sure, AI will help us make more widgets and help some people make more money.
But what if we embrace Centaur Mode not to become better cogs and commercials, but to become better at what it means to be human?
Of course, AI won’t help us figure out what it means to be human, in our own unique ways.
For instance, you’re still going to have to figure out what your High Value Activities are.
You’re still going to have to figure out what you really want so you can handle Competing Priorities, which inevitably require you to deal with some Head Trash, too.
You’re still going to have to learn to live by living.
But aren’t those activities that you already want to be doing more of? The goal of any technology isn’t to get us away from being better humans and doing more of what matters, it’s to help us get done more of the other stuff, more efficiently. AI is no different.