Ilya Sutskever, co-founder of OpenAI, who understands deep learning as well as anyone, said “I don’t know” on what the future of AI would look like. So if he doesn’t know, nobody does. Just some food for thought.
Willem Nielsen
Karma: −21
Ilya Sutskever, co-founder of OpenAI, who understands deep learning as well as anyone, said “I don’t know” on what the future of AI would look like. So if he doesn’t know, nobody does. Just some food for thought.
For years I’ve tried to succinctly formulate my issue with Effectve Altruism, this is the best I’ve done yet:
Ethical Humility: What Cellular Automata have to say about ruthless introspection, Effective Altruism, and annoyed girlfriends.
I write this to my younger self who was involved in Effective Altruism and any young people in it now.
08/02/2024 12:01am
Infinite complexity is always right around the corner. While this has been said in many ways and discovered many times, none are as concise or as simple as Stephen Wolfram’s Cellular Automaton Rule 30. Rule 30 beautifully illustrates how infinite complexity quickly emerges from simplicity and is in some ways deeper than even physics.
Rule 30 is a list of rules for transforming sets of 3 black-or-white boxes. If applied repeatedly you get a seemingly random pattern.
My version of this is what you could call the “annoyed girlfriend effect”, where a project always takes longer than you plan when hidden subproblems show up, and “just one more minute” turns into the whole evening. Put more directly, even seemingly simple problems can have huge hidden complexities.
Now when I first looked at these white and black blocks, I was underwhelmed. After all the hype I’d heard about Cellular Automata, this was it? But after thinking about if for a while, carrying this idea with me, and taking it to its natural conclusions, I realized there is quite a lot that these unassuming boxes can say.
The Woke movement for example, is based on the idea that all evil stems from the oppression of minorities. It follows that we must eradicate oppression at all costs, resulting in dangerous restrictive policies like the Canadian Bill C-16. The fundamental issue with the movement is its ideological homogeneity, meaning its unwillingness to recognize the complexity of the issue of oppression, and not account for its own ignorance. The roads to tyranny are numerous, foggy and one-way. If you consider all countries across space and time, freedom is the exception not the rule, suggesting that it is easy to land in tyranny and hard to leave. Recognizing the complexity of our world and our own ignorance that rule 30 points to, we should err on the side of caution when it comes to restrictive laws.
A similar thing can be said of the environmentalists. They are blinded to complexity by their desire for virtue.
Wokeism and environmentalism are criticized to no end but something less discussed but equally as important is the Effective Altruism movement. This is a group that aims to use science to do the most good.
In reality though it is a dangerous cult under the guise of virtue (as any reasonably good cult is).
They take a longtermist view that the majority of the people will exist in the future, therefore we should prioritize the long term. The issue is that the future has an irreconcilable amount of uncertainty surrounding it, thus the consequences of your actions are completely unknown. They claim to be open to other views but the group has become a hub of longtermism. My issue is not with their existence but with their marketing. They are not the way to virtue that many young people believe, They should really be called a hyperrational, overconfident, intellectual cult.
They miss their own ignorance, which leads them to also miss things like the value of art. More generally, doing things that don’t make sense. What comedian Theo Von calls “The Dark Arts”, or advertising guru Rory Sutherland calls “Alchemy”. Whatever you want to call it, EA misses the importance of artistry and aesthetics. The most impactful things are often the things that make the least sense, and EA people are uncomfortable with this area and avoid it fearfully, leading them to the most logical, unsurprising solutions. Why become a painter? It’s hard to give a reason, but not everyone has to have a reasonable career. In fact, that would be horrific.
Dark arts master
Their misjudgment about the complexity of the world also manifests itself in the idea that one can even get “advice” on something as personal as a career. You are constrained by your personality, your culture, your community, and so really the only way to pick a career path is to think about it for yourself (and discuss with your mother, your neighbor or your Uber driver). There are infinite variations among people and the environment, making it impossible to transfer knowledge from one person to another on questions as complex as what to do with your life.
When choosing a path there are two ways to look at it. One is you can think about what one ought to do with one’s life, and then choose paths based on that. Although useful at times, the problem with this approach is the first question is riddled with uncertainty and so you waste a lot of time trying to figure what is an ideal life (that you can’t usually achieve anyway).
An alternate approach (which EA wrongly criticizes) is to start with your constraints, looking only at the paths available to you, and then do your best to make a quick, reasonable decision on which path to take. Now that doesn’t mean you can’t think through your decisions, but you will find most decisions are not worth spending much time on and it is usually better to stick and move.
Vhinod Khosla noted this in the startup world: “Startups are a series of gut calls, you never have all the information you want and you have to just try your best.”
This constraints first approach naturally leads into doing something you enjoy (people can’t sustainably do unenjoyable things) (unenjoyable in a deep sense, not pleasurable), which of course EA criticizes.
At the heart of it, the missing piece of these movements and what rule 30 so beautifully shows is, our ignorance. It comes back to something more important than intelligence, which is honest introspection, and learning to enjoy suffering.
To boil this essay to one question, am I in this community because it makes sense, or because it feels good? In the words of Feynman, “You must not fool yourself, and you are the easiest person to fool.”