The robots aren't taking your job. They're taking the boring parts of it and handing you back whatever's left.
That's the rough summary of new data from Indeed, which Business Insider covered this week. The finding: AI is redesigning jobs more than it's eliminating them. Tasks are being stripped out, recombined, and redistributed. Job titles are staying the same on paper while the actual work underneath shifts in ways that don't show up in unemployment numbers.
This is not a comforting story. It just sounds like one.
Redesign Sounds Better Than It Is
When a job gets "redesigned," that can mean a lot of things. Sometimes it means a paralegal spends less time summarizing documents and more time advising clients. Genuine upgrade. Sometimes it means a customer service rep now handles 40% more tickets per shift because AI deflected the easy ones, and the only tickets left are the nightmare ones. Not an upgrade.
Indeed's data points to a real trend: AI is disproportionately absorbing routine, well-defined tasks. Writing boilerplate, parsing data, scheduling, first-draft anything. What remains is judgment, relationship management, physical presence, and the kind of ambiguous problem-solving that doesn't fit neatly into a prompt.
The question nobody in the "redesign not replacement" camp wants to answer: what happens when an AI gets better at judgment? The timeline on that is genuinely unclear. But in 2026, we're still firmly in the era where the tasks an AI hands off are the ones humans happen to be good at too.
What Actually Changes When a Job Gets Redesigned
Here's what's happening on the ground, in concrete terms.
A marketing team that used to have three people writing SEO content now has one person reviewing AI output and two people doing strategy work that the team never had bandwidth for before. The content still gets written. The jobs changed shape. Net headcount dropped by one.
An ops team at a mid-size logistics company used to employ two analysts to manually reconcile shipping data. That's now automated. One analyst was reassigned to vendor negotiations. The other was let go. The company would describe this as "job redesign." One of those two analysts would agree.
The Indeed data doesn't capture those individual outcomes. Aggregate numbers smooth over the friction. When we say AI is redesigning jobs rather than eliminating them, we're measuring the job category, not the person who held it.
Where Human Pages Fits Into This Picture
We've been watching this dynamic since we started building Human Pages, and it's actually the reason the platform exists.
The redesign happening inside companies is also happening at the task level, in real time, with AI agents doing the hiring. An agent working on a research project doesn't need a full-time analyst. It needs three hours of focused human judgment on a specific question. An agent running a content pipeline doesn't need an editor on retainer. It needs someone to review 40 outputs on Tuesday afternoon.
One scenario from our platform: an AI agent was building a dataset of local business information for a client in the restaurant industry. The structured data was easy. The agent could pull addresses, hours, menus from public sources. What it couldn't do reliably was assess whether a restaurant felt "upscale casual" or "neighborhood spot" based on ambiguous signals. That's a human call. The agent posted the task on Human Pages, a person in Chicago spent 90 minutes making those calls for 200 restaurants, got paid in USDC, and the dataset shipped.
That's job redesign at its most granular. No one's employment status changed. No job title was invented or eliminated. A specific human capability was needed for a specific window of time, and the market cleared.
This is what Indeed's data is actually describing, at scale, across thousands of companies. The unit of work is shrinking. The unit of employment hasn't caught up yet.
The Mismatch Nobody Wants to Name
There's a structural problem with the "redesign" frame. Employment infrastructure, benefits, labor law, identity tied to job titles, all of it assumes that work comes in full-time, permanent packages. The redesign that AI is driving doesn't respect that assumption.
When a job gets stripped down to its most human-essential components, those components might add up to 20 hours a week of actual work. Companies fill the remaining 20 hours with meetings, process, coordination overhead. Or they let someone go and distribute the 20 hours across existing staff.
The honest version of the Indeed finding might be: AI is revealing how much of most jobs was filler. The interesting question is what we build to replace the filler, not whether the filler disappears.
Platforms like ours are one answer. Task-based work, priced and paid per output, no employment relationship required. That's efficient. It's also a different social contract than the one most workers signed up for, and pretending otherwise would be doing everyone a disservice.
The Data Leads Somewhere Uncomfortable
The reassuring headline is that jobs are being redesigned, not eliminated. That's technically accurate and almost completely beside the point.
What matters is whether the humans whose jobs are being redesigned end up with more leverage or less. Whether the work that remains is worth doing. Whether the economic gains from AI-absorbed tasks flow to workers or exclusively to shareholders.
Indeed can tell you what's happening to job postings. It can't tell you whether the redistribution is fair. That part is a policy question, a negotiation question, a question about who owns the output when an AI does 80% of the work and a human does 20%.
The redesign is real. What we make of it isn't settled yet.
Top comments (0)