ph.d. in applied microeconomics, periodically thinking seriously about the impact of AI on employment and wages since Move 37.
Tim H
Karma: 196
Ah, a fellow student of Hasok Chang? I’m looking forward to digging into this.
Hmm, there may be a issue here in assuming that the market allocates capital to maximize output. Workers would be willing to pay for more human-complementing capital (via a lower wage), which would lead to a higher
Thank you. Right, it’s not about being paid per se. What I actually have in mind (but didn’t write) is more about impact, contribution, working together with others, etc. Work with all those characteristics will also be made scarce by ASI, I fear. So maybe what I want to say is: what characteristics of the work make them enjoy it, what features are decisive? And how does that set of decisive features compare with the types of work that will be made obsolete by ASI? I’ll have to keep working on clarifying this (and maybe I’ll eventually see I’m mistaken).
@Eli Tyre To clarify what I see as inconsistent, it is to simultaneously:
Support ASI displacing all paid human work because it would enable everyone to spend their time doing unpaid activities of their choice.
Currently have net worth sufficient to retire early and spend all your time doing unpaid activities of your choice.
Currently be doing paid work (that would be done instead by ASI, were it available) nonetheless—and enjoying it more than you would enjoy being retired currently.[1]
Working at a frontier lab or otherwise contributing towards the arrival of ASI is not essential to this scenario. I just imagine that many such people currently meet these three criteria.
[1] With regard to the meaning of “enjoying” here, the idea is that this criterion of human wellbeing in #3 correspond to the criterion implicit in #1′s notion of people being better off when freed from paid labor.
In that scenario, there wouldn’t be “folks who have spent decades at work on certain problems, in part based on a reasonable belief that they have a comparative advantage in their area” (as I put it in a different reply). But, right, such advocacy would strike me as difficult to justify.
If it were just one field’s problems being solved, I wouldn’t even consider such advocacy. My concern is about there being no interesting problems left for humans to solve whatsoever.
The scenario I worry about: my grandchildren are born into a world in which there are no interesting problems to solve at all, one in which there is effectively a genie who can both answer any question and make anything physically possible to make. If they are still born with a need to do meaningful work, that scenario seems cruel, and advocacy to change that status quo strikes me as reasonable—all the more so, advocacy to prevent that scenario if bringing it about has costs and/or risks. (Some may propose to instead modify my grandchildren somehow to remove their desire to contribute meaningfully with their work. My instinct is to oppose that but I haven’t given it much thought.)
We could continue paying mathematicians if we want. But what we (likely eventually) can’t do is restore their ability to make new discoveries with their minds, to receive credit for that and know their efforts contributed to knowledge. There appear to be folks who have spent decades at work on certain problems, in part based on a reasonable belief that they have a comparative advantage in their area. And within a short time, the situation is radically changed and they are transformed from an aspiring Mozart to an aspiring Salieri, only able to appreciate and perhaps help explain to others the accomplishments of AI.
What if people value their work mattering?
What work do you expect to find important or meaningful after ASI?
Parsimony should be understood as merely a heuristic for how well a model could have predicted held out data. For example, the AIC approach to penalizing model complexity in statistical modeling is asymptotically equivalent to leave-one-out cross-validation for model selection. This Stone (1977) result should be understood as an explanation for why parsimony seems to be related to truth: post hoc fit of a parsimonious model is mathematically related to how well the model could have predicted held out data. Whereas parsimony has no direct epistemic relevance, predictive accuracy is the actual goal. Since the latter is what we care about, why not just consider predictive ability directly?
you no longer fear death because you know heaven awaits you, everything is meaningful because god, and you can connect with other people over believing in god
This is not my experience of being Catholic. I’m doubtful the “old internet atheists” (or your other sources) have given you a steelmanned version of “the christian god” etc.
Hopefully it’s not hard for you to imagine someone similarly dismissive of the “glorious transhumanist future,” ridiculing it as hacky wishful thinking—because they have not taken the time to understand what you actually mean by that phrase, the depth of thought (and healthy skepticism/agnosticism) underlying it.
Option (B), then? It is not Jevons paradox, but ya’ll are helping me see there’s nothing particularly puzzling here.
The value of doing so, in this analysis, is only the benefit to others? It’s a purely altruistic motivation? Or is it not the case that they take deep satisfaction in performing this service?
I mean, the nature of much human work is expending effort to benefit someone else or, more generally, a larger project. People find doing so meaningful, sacrificing a shallow sort of comfort or pleasure for the deeper satisfaction of accomplishing something good. Does it not seem at least ironic that this particular noble sacrifice could end the possibility of meaningful work for others?
I acknowledge the logical possibility of a noncontradiction here, but I am skeptical that in reality these folks would rather be doing something other than the engaging, exciting, meaningful work they’re doing.
Tim H’s Shortform
Among people working at frontier AI labs (or otherwise contributing to progress towards AGI), how many (A) both view it as a good thing to “free people from work” and (seemingly inconsistently) continue working themselves despite adequate personal wealth? Otherwise, is it mostly that (B) such folks need/value the additional earnings, or do they (C) consider their contribution towards the loss of meaningful paid work to be outweighed by other considerations? Or do they/you think (D) plentiful decent jobs will continue persist, or is there something else I’m missing?
Today, AI solved not one, but NINE open problems – some 50 years old.
The nine Erdos problems discussed in the new AlphaProof paper were not newly announced. At least, the first I checked (125) was announced back in February.
Right, I actually read that. But is it not missing an explanation of why those mentions increased under the Nerdy personality in the first place? If the Simon Willison post (which I also haven’t seen anyone else discussing) was the origin, that seems worth noting and understanding. And both its timing and Simon’s nerdiness (in a good way) seem to fit.
update: Nevermind, apparently people were already noticing goblin mentions in April 2025, months prior to that post.
Isn’t the explanation just that an influential AI blog named GPT 5 his “Research Goblin”?
Why does GPT-5.5 love goblin mode so much they had to give twin instructions to cut out all unrequested mentions of animals? Good question.
Shouldn’t Simon Willison be the prime suspect? His prominent blog called GPT-5 Thinking his “Research Goblin.” https://simonwillison.net/2025/Sep/6/research-goblin/ And the timing fits, assuming 5.1 training included his post (and associated commentary).
update: Nevermind, apparently people were already noticing goblin mentions in April 2025, months prior to that post.
My intuition is: That line represents the point at which people start thinking “This bureaucratic structure is too cumbersome to get anything done with this many people; we therefore need strong leaders who can act through personal authority rather than merely bureaucratically-delegated authority.” I.e., the guild starts turning into a cult.
I wonder how often it would work better for guilds that grow too large to explicitly split apart. My intuition is that there should be a norm that growing guilds should split apart (and, where applicable, appoint representatives to a higher level org. to coordinate—initially a clique but later a guild if the number of base level guilds increases).
The reaction of the ML community is a textbook example of Strevens’ “iron law”, no? Until recently, there was no empirical validation of the concept of AGI, and so it was rejected by people socialized into the “iron law” norm of science. (The iron law as conceived by Strevens only admits empirical tests, not “rigorous proofs” as you include in your definition of the core scientific norm here.)
You seem to think that what is misguided is the way scientific institutions implement the iron law (requiring self-contained/justified incremental paper-sized units), whereas I (with Strevens, in my reading) think the iron law is itself poorly suited for governing short-run practical discernment (as opposed to growing the coral reef of empirical observations for long-run benefit).
An instructive recent case of the iron law hamstringing practical discernment was the bizarrely slow acceptance of COVID-19′s airborne transmission. The short version of why that happened is that scientists are not trained to step back and ask how likely a given hypothesis is, all things considered. They’re trained to dismiss deviations from the status quo default hypothesis unless there is decisive new evidence against it. (The reason for that norm, as Strevens explains, is how well it incentivizes painstaking experiments and other empirical observation.)
The reason I’m trying to sharpen this point is that I think it essential context for decisionmakers trying to sort through the he-said-she-said of disputes between rationalists, say, and academic scientists. Those refereeing these disputes should expect academics to dismiss unorthodox ideas absent decisive empirical tests. Such “iron law” enforcement serves a long-term scientific purpose but should not unduly muddle the discernment we need now.