Note that the most of the people that said lots of jobs would soon disappear were running LLM companies trying to overhype their wares.
Note further that I told you that the hype wasn’t likely true (e.g. in my essay “25 predictions for 2025” I correctly predicted that fewer than 10% and probably less than 5% of jobs would be replaced that year).
Narrative violation: AI is creating engineering jobs
Positive users endorse Gary Marcus's claim that AI creates engineering jobs by exposing hype incentives and preserving need for architects and debugging, while negative users stress LLM limits and doubt large-scale hiring gains.
No Digg Deeper questions have been answered for this story yet.
Only a narrative violation if you believed people with a lot of money at stake.
• Most of the people that said lots of jobs would soon disappear were running LLM companies trying to overhype their wares.
• Personally, I told you that the hype wasn’t likely true (e.g. in my essay “25 predictions for 2025” I correctly predicted that fewer than 10% and probably less than 5% of jobs would be replaced that year).
Narrative violation: AI is creating engineering jobs
@venturetwins it all depends on whose narrative you believed.
i said from the start that this narrative wasn’t true.
Narrative violation: AI is creating engineering jobs
I can admit AI is creating Software Cleanopers
@GaryMarcus I think it will be mixed. Some companies will still try to shoehorn trash AI into their workflows just to save money.
Similar to how they all outsource customer support even though it's terrible and every customer hates it.
But on some tasks, they'll give up on AI.
@GaryMarcus Claude, clean up your slop. Don't make any mistakes. Boom done ✅
@behradkhodayar @GaryMarcus I would call it real software architects getting hired again.
@GaryMarcus The hype around AI has most definitely deterred many from getting into software engineering. This plus the tsunami of slopware will result in actual software engineers being in higher demand.
AI didn't remove the engineering work — it relocated it.
Producing code got cheap. Understanding the system, debugging it, making it hold together did not. That cost doesn't disappear; it gets deferred downstream and re-priced as refactor and consulting work.
"AI replaces engineers" confused the cost of typing code with the cost of shipping working systems. Never the same line item.
@GaryMarcus The incentives are backwards. LLM companies push job loss fears to hype. The actual shift in markets? Much smaller. And weirdly, that's where the real action is.
@GaryMarcus the whole "jobs will vanish" story being pushed by people selling the thing makes it so easy to spot
glad someone said it out loud
@Jlafortetech @GaryMarcus As long as customers don't abandon the service due to a decline in quality (such as AI translation), they will continue to use "AI".
@GaryMarcus Exactly. AI speeds up code output, not systems thinking. Refactoring, architecture, and reliability still need engineers.
@GaryMarcus hey diem me plz
@FProlog @behradkhodayar @GaryMarcus But not many. If I can for example clean up 10 codebases in a month, each over 100K LOC, how many of me would need to be hired? Do you think we don't have AI tools to help with refactoring, cleanup, and cybersecurity?
@behradkhodayar @GaryMarcus But when one person can do the job of 100 people, how many jobs is that going to create? One person can clean up 10 codebases and with enough compute, do it relatively quickly, in weeks.
@GaryMarcus The loudest job-apocalypse voices had the most to gain from the panic. Incentives matter when reading predictions.
@LexTweetr @GaryMarcus Shame you put "don't make mistakes" in the middle. LLMs struggle with the middle of instructions.
@GaryMarcus Coding was meant to be the GenAI use case. A few trillion bucks later and we've got one LLM that can sort of code, and a bunch that simply create new human work.
Economic growth: unlocked by AI.