The age of the subconscious
Greetings from Göttingen in central Germany,
I decided to jump on a train and work from this lovely university town this sunny Friday. It’s a place of rich nerdy history as it was the former workplace of mathematicians Gauss, Riemann, and Hilbert
The past few weeks have been busy with creative work around the theme of storytelling. I published a few Youtube videos (if you like street art, you might like my visit to Berlin in search of creative inspiration), and I have published my first short story where a complex father and daughter relationship comes to a fork in the road.
Less time was allocated to my other activities like coding and investing. And I believe this is a manifestation of a more general feeling I have had recently in relation to AI developments: the domain of the logical reasoner may not be where we will make an impact or find fulfillment anymore. Let me explain.
The roll-out of new AI technology in the past six months has ignited deep questions in all of us about our role in society and how we will work and live. For me, cooperating with GPT4 on coding or writing work has triggered some thoughts about the fundamental characteristics of being a human.
For most of my life, I have treasured rational reasoning. To me, being smart was a means to rise above my initial situation in life. I believed that knowledge was power, the power to influence my future.
To some extent and with some luck, this has panned out as expected. Intelligence and hard work have been the key traits to succeed in the capitalist order.
But we seem to have reached peak Homo Sapien, the “Wise Man” defined by his/her brain compared to other species.
Why do I say that we have reached a peak? Because I have seen this thing, GPT4, do reasoning work at the speed of light. And I cannot imagine what the next iteration will be capable of doing.
Whenever I code a little program, it takes me a few hours, and I may have to overcome five to ten errors, which GPT4 helps me with. I just can’t help but think that in a couple of years this will probably take a few minutes and maybe will be done automatically by the AI. My motivation to master this activity diminishes. I still want a level of proficiency to accomplish outcomes, but the continuous learning path of adding knowledge cumulatively to accomplish something is almost breaking down.
We tend to say that this shift will free up time to do higher-order stuff or that it can allow us to leverage our impact in fields of reasoning and thinking. “Software engineers will become 10x engineers”. This may be the case in the short term. But I don’t think there will even be a contest on the rational reasoning side between us and AIs. How could there be one when AIs can access all written and visual content on the internet and some of the conceptual knowledge used to create it?
We will likely delegate the pure logic and reasoning task to the AIs. My vision is that humans will be initiators and providers of hypotheses but that they will delegate much of the heavy lifting to the AIs.
What does this mean? Let’s put it on a longer timeline. I’m pretty confident that no human being will write code in 10 years. I’m also confident in a less common prediction: no human being will teach any factual-based knowledge.
Why? Because it is very plausible that AIs will be much better at this than us. By that stage, they will make no errors and customize their behavior to the humans they assist. These activities will not be the domain of humans anymore. Just like calculating things is not the domain of humans but of electronic calculators and traditional computers.
Retracing our steps to a nearer future, say the next couple of years, what are the implications? My firm belief is that the internet will be flooded with factual AI content of the type that teaches people something, be it a blog post, a high-brow essay, or a Youtube tutorial. And I believe that humans will prefer AI content to human content in these fields. Just like I prefer to fly on an airbus a320 with autopilot than some of those small old planes with propellers used on low-traffic routes.
To me, writing the types of essays I was writing a few months ago, deep-diving into an idea in a cold and impersonal manner, will be as enjoyable as writing down a multiplication table or the solutions of cubic equations. People used to do that manually and enjoyed it. It was part of the mathematician's identity in the middle ages, in fact. Not anymore.
The physicist Max Tegmark suggested that we should probably rebrand ourselves. Here is the time stamp at which he talks about this in a recent discussion with Lex Fridman (watch it for 1-2 minutes)
I would go further than Tegmark and say that it is not just our awareness and subjective experience but also the subconscious processes in our minds that will feature more prominently going forward.
These two sides of our mental abilities are the hidden factors that bring us art, creativity, and imagination. They also help us go above our preconceived limitations.
Most athletes highlight that above a certain level of skill, mental factors help them achieve higher and higher levels of performance. Much of it is conditioning the subconscious to go over a pain threshold, stay calm and not panic when things get tough, or simply trust our muscle memory.
These are not mental processes you can bring forward with reasoning. They seem to be acquired over a period of time in an unplanned way. An analogy I like is shooting in basketball. Passed a level of technical training and repetition, it’s really the belief that your muscle memory will do the right thing in the few seconds between receiving the ball and shooting that matters. Great shooters like Steph Curry almost don’t think. They try to get into the groove and let their subconscious processes and skills take over.
This AI technology, if utilized well, could allow us to live in a world where people are busy producing things that affect other people and that only humans can do. It is a type of knowledge we have not codified explicitly.
I was reminded of this while recently reading a book about short stories and art.
Say the painting has in it that healthy tree and a second one that, on first glance, looks identical. The mind immediately starts scanning for differences. Say there’s a bird in one of the trees, barely noticeable. Now we read the bird-containing tree as “life-welcoming,” the other as “barren.”
We are always rationally explaining and articulating things. But we’re at our most intelligent in the moment just before we start to explain or articulate. Great art occurs—or doesn’t—in that instant. What we turn to art for is precisely this moment, when we “know” something (we feel it) but can’t articulate it because it’s too complex and multiple. But the “knowing” at such moments, though happening without language, is real. I’d say this is what art is for: to remind us that this other sort of knowing is not only real, it’s superior to our usual (conceptual, reductive) way.
George Saunders, A swim in a pond in the rain
In a way, the changes we are experiencing are helping us reflect on what we are. That’s healthy and may lead, in the end, to more compassion towards all beings that don’t exhibit intelligence, whether human or animal, but share those indescribable notions of being aware, sentient, and knowing some things almost subconsciously.