Like the kid with the spoon in The Matrix, it’s time to realize that … there is no lesson. At every turn of the year, like many others, I have offered my lessons from the previous twelve months. It’s a futile ritual. We never learn.
We never learn
A few examples, starting with economics: Inflation, which had surged to the highest rates in half a century, is finally coming back to less disruptive levels. It’s taken two and half years, through which US inflation averaged 6%, and an extremely sharp monetary tightening. Now hopefully the job is (almost) done, and without a recession.
After a well-deserved sigh of relief, we could look back at this painful experience in two different ways.
Option A: “well done central bankers, but it was needlessly reckless to run extremely expansionary monetary and fiscal policies for a decade. Luckily we’ve restored stability with limited damage (though consumer prices are nearly 20% higher than at end-2020), but now let’s calibrate macro policies on a more sensible setting so we don’t have to go through another crisis soon.”
Option B: “see, it was just a transitory shock, now we can go back to extremely low interest rates and ultra loose fiscal policy. Let’s watch that Modern Monetary Theory video on YouTube again.”
We’ll take Option B, of course.
Politics and geopolitics are rife with examples of lessons not learnt:
Exhibit A: the US Presidential election is careening blindly towards another Biden-Trump showdown, again with alarmist warnings that democracy is in peril.
Exhibit B: geopolitical sparring and fighting has brought us back to the good old 20th century, with Russia still trying to occupy Ukraine, Iran becoming increasingly confrontational while it builds its nukes, and China flexing its muscles.
Exhibit C: speaking of China, after enjoying a remarkable run of strong growth thanks to some good economic liberalization, the country that claimed to have learnt from the Soviet Union’s mistakes now seems intent on reestablishing a tight centralized grip on the economy.
Does it look like we’ve learnt anything?
We don’t want to learn
As much as we seem to be obsessed with learning lessons (articles and books on “The ten lessons of …” enjoy outsized popularity throughout the year), we seem less and less interested in learning — at least in the West:
Declining learning scores attest to a steady deterioration of school systems;
Universities prioritize social and political indoctrination to the detriment of open debate and critical thinking;
We crave for AI to provide pre-cooked answers to all questions.
So let’s face it, we don’t really want to learn; we want easily available answers so we don’t need to think. That’s how we have evolved into a society of polarized echo chambers: choose your team and it comes with a full set of pre-approved certainties on any relevant issue: from economics to immigration to geopolitics to the media. Much easier than doing research, examining data and listening to different opinions with an open mind.
Maybe we can’t learn
The defining event of 2023 was the Hamas terrorist attack on Israel. The acts perpetrated on that day were unimaginably horrifying, the accounts harrowing. And yet they have been followed by a shocking, open resurgence in antisemitism; Western societies obsessively careful not to offend anyone with a wrong pronoun have taken to chanting the most threatening racism, and feeling all the more virtuous for it. Maybe we can’t learn, maybe we’re just not wired for it.
Learning engines
Maybe machines can do the learning for us.
Artificial Intelligence has been all the rage in 2023 with the rapid spread of Large Language Models like ChatGPT. Half of the AI community confidently predicts that Artificial General Intelligence is just around the corner, while the other half cautions these LLMs are nowhere near as revolutionary as they are made out to be.
I side with the skeptics: there is no evidence yet that LLMs “understand” the world in the sense of being able to generalize what they learn and distill it into laws and insights that can lead to innovative problem solving. And I do worry that the first applications of these models will be overwhelmingly harmful, inundating us with misinformation and deepfakes, indulging our lazy appetite for easy answers and further eroding our cognitive abilities.
But they do force us to think harder and deeper into what intelligence is. In terms of harnessing the existing stock of knowledge into coherent and persuasive answers, LLMs are already capable of very human-like feats (including their unfortunate tendency to hallucinate and make stuff up.) We can brush it off as statistical prediction, but we can’t rule out the possibility that this is getting us closer to how our own intelligence works. Maybe AI will prove better at learning than we are.
Ethical Learning
Can we lead AI to learn the right moral values, so that if it becomes sentient and self-determined it will be “ethical”? Asking this question might be a waste of time: If AI achieves true AGI level, it’s foolhardy to believe we can influence its goals and ethics. If instead it will just be a powerful supercomputer, the question is moot.
More important is that we cannot agree of what is ethical. Think of the ongoing debates on diversity equity and inclusion, the Israel-Hamas conflict, freedom of speech. Even within the framework of Western culture we can’t agree on what is ethical; and other cultures (think of sharia law) have dramatically different frames of reference. Whose ethics do we want to the AI to adopt?
The experts who purport to champion ethical AI keep warning us that AI might lead to human extinction, yet keep beavering away at it under the greedy watch of tech and media giants. Rather than worrying about ethical AI, perhaps we should worry about (un)ethical humans. And hope that AI proves to be a more ethical learner than us.
Maybe we can learn to learn
Let’s be optimistic. Maybe we can still learn to learn. The main lesson (!), though, is to recognize that learning does not come easy to us humans. So let’s be realistic and focus on just a few important issues. And let’s not trust the easy answers – not even from the experts. Let’s just think.
I expect that a year from now I’ll be back with a “Seven Lessons For 2025” post — I never learn either.
😀 an 'open mind'? What on earth is that?
Ah! I must be too young to remember!😀