The 56% wage premium makes sense but masks a bigger gap. Most "AI skills" on job postings mean "can write prompts." The real premium - the one that'll widen - is for people who build with AI daily.
I've been running an autonomous agent for months. Not a side experiment - actual infrastructure that manages tasks, deploys code, works night shifts. The distance between "I use ChatGPT" and "I built a system that fixes itself at 3 AM" is enormous.
The problem is that the hype is not just hype. There's truth there and that's why companies want those skills in the first place. Whether the extent of their value correlates with the hype is another question but not as important in that sense
I'm not saying there's no basis for the hype- I wish the public in general would take the prospect of AI job-replacement more seriously, not less. I'm trying to point out there is no way to "leverage the technology to enhance your career" unless you literally work for Anthropic or DeepMind or something. There is going to be a small-to-nonexistent time-horizon between "it replaces the junior devs" and "it replaces the senior devs" and "it replaces the project managers" and "it replaces the CEOs".
You can ban AI, or you can be ruled by it, but as an individual you cannot 'exploit' it outside of that vanishingly small time window. Unless the researchers in the Altman Brat Pack hit some kind of wall, either technical or political.
You assume it's "vanishingly small" but I disagree. The more you learn AI skills, the larger that window is for you. That's the whole point! You're mixing the idea of AI as a tool with the idea of AI as a super intelligent being and that's a dangerous place to be in because it's paralyzing (and, I think, misleading)
Unless you think the delay is going to be measured in centuries I really don't see how it changes the prescriptions here. We either need to ban the technology or resign ourselves to being, essentially, hominid housepets in a robot civilisation subsisting on UBI allowances. (And in the latter case, I'd like to know when the citizens of Brazil or France get a cheque in the mail from tech companies with patent moats based in Shenzhen or San Francisco.)
Or, I dunno, I guess the bubble could pop first and we get an AI winter instead, around the same time the OECD's pension systems start to collapse.
A career in big tech left me survival-coded to run to the next big thing. There have been lots of next big things and through that lens, AI is no different. My buddies still at MSFT are generally thrilled to have AI around to do their bidding; there was never enough hours in the day to get the job done. The ones that got RIFed probably don’t feel the same.
The 56% wage premium makes sense but masks a bigger gap. Most "AI skills" on job postings mean "can write prompts." The real premium - the one that'll widen - is for people who build with AI daily.
I've been running an autonomous agent for months. Not a side experiment - actual infrastructure that manages tasks, deploys code, works night shifts. The distance between "I use ChatGPT" and "I built a system that fixes itself at 3 AM" is enormous.
Documented what that learning curve actually looks like here: https://thoughts.jock.pl/p/wiz-1-5-ai-agent-dashboard-native-app-2026
The market will figure out this distinction. Question is how fast job descriptions catch up.
I think the market already makes this distinction. Writing prompts half well is the bare minimum!
100% but the better models are, it will become less important imho
If the hype is real, then believers will also be unemployable on marginally longer timescales.
The problem is that the hype is not just hype. There's truth there and that's why companies want those skills in the first place. Whether the extent of their value correlates with the hype is another question but not as important in that sense
I'm not saying there's no basis for the hype- I wish the public in general would take the prospect of AI job-replacement more seriously, not less. I'm trying to point out there is no way to "leverage the technology to enhance your career" unless you literally work for Anthropic or DeepMind or something. There is going to be a small-to-nonexistent time-horizon between "it replaces the junior devs" and "it replaces the senior devs" and "it replaces the project managers" and "it replaces the CEOs".
You can ban AI, or you can be ruled by it, but as an individual you cannot 'exploit' it outside of that vanishingly small time window. Unless the researchers in the Altman Brat Pack hit some kind of wall, either technical or political.
You assume it's "vanishingly small" but I disagree. The more you learn AI skills, the larger that window is for you. That's the whole point! You're mixing the idea of AI as a tool with the idea of AI as a super intelligent being and that's a dangerous place to be in because it's paralyzing (and, I think, misleading)
Unless you think the delay is going to be measured in centuries I really don't see how it changes the prescriptions here. We either need to ban the technology or resign ourselves to being, essentially, hominid housepets in a robot civilisation subsisting on UBI allowances. (And in the latter case, I'd like to know when the citizens of Brazil or France get a cheque in the mail from tech companies with patent moats based in Shenzhen or San Francisco.)
Or, I dunno, I guess the bubble could pop first and we get an AI winter instead, around the same time the OECD's pension systems start to collapse.
A career in big tech left me survival-coded to run to the next big thing. There have been lots of next big things and through that lens, AI is no different. My buddies still at MSFT are generally thrilled to have AI around to do their bidding; there was never enough hours in the day to get the job done. The ones that got RIFed probably don’t feel the same.