Most Workers Prefer AI for Repetitive Tasks, But Startups Focus on Creative and Strategic Automation

A recent Stanford study suggests AI startups are spending huge sums on tools that don’t actually solve problems workers care about. After analyzing the work preferences of people across more than a hundred professions, researchers found that nearly half of the automation being built targets tasks employees would rather keep.

The study, based on responses from 1,500 workers and mapped against a database of common job duties, shows a clear mismatch. While employees ask for help with boring, repetitive tasks, many AI startups are aiming higher up the value chain, building tools for strategy, creativity, and management. These are areas where most people would rather stay involved.

Workers Want Help With the Boring Stuff

Across the board, workers say they’d rather use AI to get rid of the repetitive parts of their day: sorting receipts, fixing basic tech issues, pulling reports, and entering numbers. These aren’t glamorous tasks, but they’re the ones that slow things down.

In one survey, more than half the respondents said they’d like help with data entry, IT troubleshooting, and building spreadsheets. Another study found that office employees spend more than 10% of their time typing the same information into systems and over half their day working on documents that often go through endless revisions.

But the tech being built doesn’t reflect that. Instead, a lot of energy is going into automating creative thinking, marketing, and planning, areas people don’t want to give up. Some founders may think they’re saving time, but for workers, it feels like the wrong trade-off.

The Funding Frenzy Isn’t Helping

Startups follow the money, and right now, that money is chasing big promises. In early 2025, over $73 billion in venture funding went to AI startups. That’s more than half of all VC investment in the quarter. But with that flood of cash came a wave of flashy prototypes and rushed launches. Instead of building tools that quietly fix daily frustrations, many companies focused on eye-catching demos and futuristic ideas.

That approach has consequences. Recent reports show that nearly half of all AI projects inside companies end up being abandoned. The tools don’t meet real needs, or they’re too complicated to use. In the worst cases, they create more work than they save.

People Want a Partner, Not a Replacement

Digging deeper into the Stanford data, one theme keeps coming up: most workers don’t want to hand off everything to machines. They’d rather work alongside AI, not beneath it. The study introduced something called the Human Agency Scale, which measures how involved people want to be when AI is in the mix. Across nearly half of the occupations studied, the most common preference was right in the middle, a balance between machine assistance and human control.

That preference wasn’t just philosophical. It reflected how people actually feel about their work. They don’t mind letting AI take care of the chores, but they want to stay involved when it comes to judgment, creativity, or communication. In short, they want support, not replacement.

Trust Is Still a Big Problem

Even when AI tools are useful, plenty of workers don’t trust them. Some worry about mistakes. Others don’t believe the systems are reliable enough. And a lot of people are still concerned about losing their jobs.

That unease shows up in the numbers. Around 45% of workers say accuracy is their biggest concern with AI tools. About a quarter worry about being replaced. And more than 15% feel that handing over parts of their job makes the work feel less human.

These feelings can create pushback. In one report, nearly a third of employees admitted they’ve actively avoided using company AI tools. Younger workers, in particular, were more likely to resist. Some said they’d expected more from the tech and were disappointed when it didn’t live up to the hype.

The Real Wins Are Hiding in Plain Sight

It’s easy to overlook the small stuff, but that’s where AI could have the biggest impact. Every day, office workers lose time to tasks that don’t require much thinking, updating forms, copying data between apps, or chasing down approvals. For a team of 20, those tasks can add up to thousands of hours each year.

Even modest improvements in these areas can lead to real savings. Less time spent on admin means more time for meaningful work. And unlike high-level automation, these changes don’t threaten anyone’s role.

A Smarter Approach Starts With Listening

The lesson from the Stanford study is simple. If AI tools aren’t designed with workers in mind, they probably won’t get used. The companies that succeed will be the ones that ask what people actually want help with, then build for that.

That might mean setting aside the idea of revolution and focusing on small, steady improvements. It might mean designing interfaces that are easy to trust and letting users stay in control. And it definitely means bringing workers into the process, not just handing them a new system and hoping it sticks.

For investors and startup founders, this is a moment to rethink priorities. Big promises might grab headlines, but the real value often sits in the tasks everyone hates and no one has automated yet.

Nearly half of current AI tools target the wrong problems. Fixing that starts with a simple shift in mindset: stop asking what AI can do and start asking what people actually need.


Notes: Image: DIW-Aigen. This post was edited/created using GenAI tools.

Read next: TV shrinks, podcasts grow, print fades: What the numbers say about US media use
Previous Post Next Post