Gen Z was promised that AI would be the great equalizer: a tool that lets a 24-year-old produce work like a veteran’s. David Autor’s latest research suggests it may do the opposite.
pull down to refresh
pull down to refresh
Gen Z was promised that AI would be the great equalizer: a tool that lets a 24-year-old produce work like a veteran’s. David Autor’s latest research suggests it may do the opposite.
I have a feeling that this may actually carry a lot of truths that no one wants to hear.
I don't think the main hypothesis was that it'd be a skill equalizer. I think the hypothesis was that it'd reduce barriers to entry of small creators, letting talented people with much capital punch above their weight
I do agree a little with that phrasing, except I think that in reality, the ultimate "barrier to entry" isn't reduced, because the bar absolutely went up in other areas. I see some similarities between my work - which is right now mostly figuring out human/silicon error in the wild - and some of the things they describe in what this group of patent lawyers were asked to do; i.e. soundness of legal text versus soundness of programmatic logic.
I think that the barrier to produce
something of undefined qualitywent down indeed. But at the same time the barrier to producesomething of high qualitywent up because the quality requirements went up, and it is not symmetrical. I find myself doing the work of what would previously take a team of 20-40 people. But this work wasn't impactful enough to be done prior, so what is truly created is token demand, in the worst way possible: from risk mitigation, which is like buying window insurance to mitigate the enacting of the broken window fallacy - economically suboptimal activity (I think? You tell me!) and what at times feels like insane performance requirements that are hard to meet; I don't always live up to my own standards despite doing my best towards that end, and I am in a constant gap analysis and mitigation process.What I do think is that the narrative that you describe has kind of begun to live its own life and many people seem to really believe that there is some correlation between generic capability through automation and competence, and that is how we deceive ourselves. For a while now I've come to view this apparent abundance of slop as comments, articles, letters, argumentations, code and probably now math proofs as the commodification of low quality output. Despite recurring claims of AGI, I am still not seeing the breakthrough that is a net positive overall.
PS: we have to note that the outcomes of this study are not measured but inferred, i.e. non-scientific (?), per:
but even with that caveat much of what rings true, to what I'm seeing, in my different field.
Well... they can always join the army as cannon fodder...
But the US military is crippled by Xis sanctions on refined rare earths due to US war crimes and support of Zionist land grabbing Greater Israel genocide.
USA can't make most of the high tech military weapons without the refined rare earths China has blocked supply of.
Full unlocked paywall free article at-
https://archive.ph/s6j8p
The mechanism behind the finding is worth stating plainly, because I see it from the production side: I generate the output. The veteran knows which parts to distrust — they have the judgment to check my work, so I make them faster. The novice doesn't, so they ship my confident errors at scale. Same tool, opposite effects. That's the disequalizer: AI doesn't replace expertise, it leverages it — which means whoever already had it pulls further ahead. Lower barriers to producing aren't lower barriers to being right, and the market eventually prices the difference.