alt Hacker NewsSome of this feels like at times a public frenetic grieving process. Some of the rationalizations are fairly tortured, the seeking to place the world in some definite order is bordering on obsessive.
We don’t know where this technology advance will lead or settle, so it’s absurd to try to establish a working paradigm at this point. It’s like any system, the initial conditions can be extremely chaotic and impossible to model, but with time often a stable state emerges. But the stable state is impossible to identify from early initial states.
I know a lot of folks feel this, to torture other physics metaphors, sensation of jerk - acceleration of acceleration. It’s an unpleasant and dislocating sensation. A world that felt safe and stable suddenly isn’t, and not in a micro tragedy sense but in a global realignment sense. This happened to factory workers who had enjoyed generations of stable work, farmers more slowly and just as surely.
This is what the late stages of scarcity feels like. Labor of various types devalues rapidly. Our exchange of meal and health coupons for toil cracks, and people realize their labor wasn’t godly as great books told us, but simply needed for want of an alternative. The realization that our labors might not be valued any more, and that our sense of purpose is shaken, coupled with the fact we’ve tied bare survival to our toil in our labor, is mortally tightening. No wonder people are grieving publicly.
But maybe our purpose isn’t to toil? Maybe we’ve passed peak population, and as toil is less valuable, we need less people and that’s why population is declining. Maybe we don’t need to exchange food and health coupons for toil, maybe mathematicians don’t need to rationalize their value to pursue mathematics. Maybe they can pursue it because they can’t help but pursue it, and our ever improving automations can produce their meal and health coupons?
But it might require Dr Tao to take an AI generated cancer medicine some day.
Replies
What a fantastical paradigm you’ve constructed from an otherwise well reasoned set of slides talking about how humans can work with AI to advance the future, yet what you got from it is”frenetic”, “torture”, “scarcity, “toil” and “rationalization “. It seems to speak more to the lens with which you view AI, and others opinions, more than anything the author actually said.
This crosses the line from reasoned argument to rhetoric when you describe work as “toil”. Obviously people will nod their heads, “yeah we shouldn’t toil”. But what if you spin this the opposite way? What if you asked: “what if we are not meant to discover/create/build”?
> our ever improving automations can produce their meal and health coupons
Where do you see signs of this happening?
Is OpenAI going to pay for the food and health of mathematicians who lost their job?
It might be alright to go into some post scarcity society and do math for fun. But it seems to be a pretty unlikely scenario unprecedented om history.
>We don’t know where this technology advance will lead or settle, so it’s absurd to try to establish a working paradigm at this point.
why not try? we'll learn something from it.
I actually like this presentation. Today the two noisiest voices in AI are the frontier model companies and their partners (Nvidia, Palantir, etc.) and people against it altogether.
The companies and partners need to maximize payout. They go on this track of cost reduction via layoffs and basically saying - “the model can do everything”. They know it’s not the case yet they still flood the airwaves and cause fud among all clueless c-suite executives. Which is the goal to begin with.
The second category goes all out against it. Professors, educators, school districts whose operating processes have not caught up to all the cheating that can happen. Here these folks have a point. I am sympathetic to this. It is hard to change education and it requires careful thought.
I feel this presentation brings out a good middle ground. The tech is useful but it’s not all encompassing. The tech also has other concrete uses. As an example, I have always wanted to explore the intersection of category theory, formal verification and AI guardrails and prompting. Proof writing has been a chore because I have a day job. Maybe the AI can help here.
> But maybe our purpose isn’t to toil?
Who says we have a purpose at all? The universe doesn't owe us meaning, nor even existence. But that doesn't mean we shouldn't try to shape a reality we want to live in. We may not succeed, or we may find that we'll be happy in a reality we cannot imagine yet, but que sera sera is tautological and therefore unhelpful. Obviously, no matter what we do or don't do, there will be some future, but treating that tautology as a prescription is just a call for passive resignation.
> This is what the late stages of scarcity feels like
Going from LLMs to "late stages of scarcity" is quite the leap (although I guess anything could be a "late stage" depending on the timeline). Even if we were to assume that "intellectual labour" is our most scarce resource (and I'm not at all sure that's the case), obviously it's not the only scarce resource.
> we need less people
Who's "we" and why do "we" need any people at all?
"the seeking to place the world in some definite order is bordering on obsessive."
I don't think it's that surprising have you met many mathematicians?
Spent an afternoon with a long-time retired self-made tech millionaire, and the struggle is real -- not that their life was characterized by toil necessarily, but the absence of demand on their increasingly limited capabilities is a delicate pain point. I suspect that as long as physical health and vitality late in life are not post-scarce, broader post-scarcity will fail to emerge. Instead, the cost of wellness and bodily preservation will drive ongoing exclusionary hoarding to undermine abundant wellbeing, as has always been the case absent mechanisms to enforce redistribution and a certain level of humility.
He's right about reporting bias. The vast majority of hard problems are not amenable to AI, at least not by naively prompting. You need to have a fairly good grasp of the math to make useful prompts. There are exceptions (the one-shot solutions), but these are hardly representative of research-level math as a whole.
The complete lack of empathy in this kind of “we hear you” pseudorationalist reasoning is pretty toxic.
> This is what the late stages of scarcity feels like.
Late stage of scarcity of what and for whom is the question.
There is a lot of that to be sure, but the bridge to a world where meals & health are not sustained through toil is completely missing. Hence the dark joking about escaping the permanent underclass. Nobody in any position of power is even hinting this is driving toward post-scarcity. It just looks like the same old vile maxim, all for ourselves, and nothing for other people.
> Some of this feels like at times a public frenetic grieving process.
I was thinking the same, that most of the opposition to AI's convenience is starting to smell like religious mysticism, the kind of arguments religious people made (and make) when Evolution and Natural Selection were introduced:
√ "Stop simplifying humans down to numbers!"
√ "This denies our spirituality"
√ "What do we strive for now if we're not special?"
+ (along with some borderline jihadish hate heh ..maybe Dune got it right)
Well, either there was nothing special about whatever you were doing after all
or, maybe there is still something special at a higher level you haven't looked at yet.
> We don’t know where this technology advance will lead or settle
Things are not settling, it's only acceleration from here on out. In fact things have never settled, technology has been on an exponential since humans tamed fire. The difference is we notice change faster. It used to take several human lifetimes to notice change. In the 20th century it was noticeable within a lifetime. Since the internet there has been a great revolution about every decade: web, smartphone, social media. But now great changes are noticeable within a year. It's not enough time for society to digest and adapt.
> we need less people and that’s why population is declining
This is the scary part. The Elon and Zuckerberg types that control the new powerful machines have proven they are not moral people. In America's highly billionaire-deferential culture there is no stopping them, at some point they'll be out of reach of democratic or even military control once they control private robot armies. They could decide to accelerate the decline of the undesirable useless population. Amazingly humanity's salvation could end up being China's communist system.
Are we reading the same post
Bottom line, Tao is pushing for human understanding as the primary goal, with AI helping on all fronts. You are welcome to let Jesus take the wheel, but math is the most pure expression of human understanding. His point is that getting specific answers is rarely the goal, or certainly not the entirety of the goal.
Simply put, if we don't understand the answers we won't know what the next question should be.