Understanding Pro-Worker AI in Your Career

There’s a narrative about artificial intelligence that dominates every conversation right now. You’ve heard it constantly speak about how AI is coming for your job. Machines will automate your work. The future is jobless, or at least precarious. You’ve probably heard it. I had, too, until I read a forty-page research paper that reframed the entire conversation for me.

The paper is called “Building Pro-Worker Artificial Intelligence,” authored by Darian Asdumego, David Autor, and Simon Johnson. And their argument is simple but profound: there’s another direction AI can take. One that doesn’t replace workers, but makes them better. It speaks on how the use of AI constructively can help expand labor market opportunities and increase the value of labor itself.

They call this “pro-worker AI,” and they define it as technology that makes human skills and expertise more valuable by expanding workers’ capabilities. Not replacing them, but again expanding them. That distinction matters enormously, because it’s what I want to break down with you here.

The Two Stories AI Can Tell

When most people think about AI and work, they think about automation. And fair enough, that is understandable. It was our real AI before there was AI. But if we are speaking truthfully, AI has genuine capacity to automate tasks, displace workers, and reshape entire industries. And, the research acknowledges this directly.

But here’s what doesn’t get talked about enough; AI’s capacity to act as a force multiplier for human skills and expertise is equally substantial.

Think about it this way. Automation requires rules. In order to make those rules, you have to know what you are doing manually. The focus then shifts on break down each task into discrete steps, codifying them, testing them, and then pushing them to run. It’s binary yes or no, do this or do that. Automation works when the problem is well-defined and repetitive.

AI on the other hand operates differently. It doesn’t need perfect rules because it learns from patterns. It can handle ambiguity. It can ask questions. It can coach you through a problem you’ve never solved before. This is where the opportunity is.

Its funny, two years ago, my company started pushing cloud adoption. A year later, they pivoted hard into AI. I was already using it, mostly because I had questions nobody around me could answer. I had vision, I could see what we needed to do but, I didn’t have direction. My boss didn’t have the answers either. So I started asking AI. And something shifted, instead of feeling threatened by AI, I felt amplified by it. I could now use AI to help me along the way of learning a new skill.
That’s pro-worker AI in practice.

What Pro-Worker AI Actually Is

The authors make a critical distinction that most people miss. They argue that “rather than seeking to automate expertise entirely, delegate decisions wholly to machines, or assume either humans or AI will invariably excel, the objective should be collaboration; integrating human judgment with rapidly advancing computational capabilities to achieve superior outcomes.“

Pro-worker AI doesn’t promise to replace your expertise but instead makes your expertise more valuable by extending its efficiency and scope. It’s the complementarity between machine capacity and human expertise that creates the pro-worker potential.

Here’s what that looks like in practice: an expert human with domain knowledge partners with AI. The human brings judgment, context, ethical reasoning, and the ability to navigate ambiguity. The AI brings speed, pattern recognition, access to vast amounts of information, and the ability to simulate scenarios the human has never encountered. Together, they’re better than either could be alone.

The paper gives concrete examples. Aircraft maintenance technicians use AI assistants to troubleshoot problems faster. Patent examiners use tools like “More Like This”, software that suggests relevant prior art documents, to conduct more precise searches and reduce examination time. Teachers could use AI to build curriculum, grade assignments, and identify struggling students, freeing them to focus on what humans do best: mentorship, inspiration, and connection.

In each case, the technology made the human’s expertise more valuable.

Why This Matters for Your Career

Here’s where this gets personal. The research identifies five broad categories of how technology reshapes the value of skills and expertise. Most of them: labor augmenting, capital augmenting, automating, are morally neutral. They can go either direction. But one category stands out: new task-creating technologies. These are “unambiguously pro-worker” because they create entirely new types of work that didn’t exist before.

Twenty years ago, data scientist wasn’t a job. Now it’s one of the highest-paid roles in tech. Why? Because digital transformation made that expertise newly relevant and valuable. The technology didn’t eliminate jobs, it created demand for entirely new expertise.

Right now, that’s happening again. Prompt engineering. Context engineering. AI governance. These roles barely existed two years ago. They exist now because AI created demand for them. And that pattern will repeat, because new problems always emerge.

Here’s the catch though, most people don’t realize this is a choice. When you automate a task, you’ve freed up time. The question is what you do with that time. Do you sit in the same role doing the same thing, just faster? Or do you ask, what’s the next problem I can solve?

I’ve noticed something working in privacy and governance. Companies will automate you right out of a task but they won’t automatically redirect you somewhere else. That’s on you. You have to be willing to learn, to ask questions, to move into ambiguity. And the people who do that? They become indispensable. They become the ones who understand both the old way and the new way. They become the bridge.

Why Pro-worker AI isn’t everywhere

The authors identify why pro-worker AI isn’t everywhere yet, and it comes down to misaligned incentives. There’s a difference between the people who buy AI and the people who build AI. Developers and vendors are laser-focused on cost reduction. Their customers want automation. So that’s what gets built.

Pro-worker AI, the kind that augments instead of replaces, is harder to sell. Especially if you do not have a use case nor understand your workers’ actual needs. It requires designing for collaboration, not substitution. It requires thinking about learning and skill development, not just efficiency.

There’s also the human side. Workers often resist tools that require them to learn new things. Comfort is powerful. Stagnation is easier than growth, but here’s the opportunity. There will always be demand for people willing to grow. There will always be gaps that need filling. The question is whether you’re filling them or watching someone else do it.

The other obstacle the research points to is workplace surveillance masquerading as worker support. Some companies use AI monitoring tools that track real-time audio, video, even worker movement to “enforce standards” at this current moment. That’s not pro-worker. That’s dystopian and prison. And as someone certified in privacy, I can tell you it creates legal, ethical, and cultural problems that no efficiency gain is worth.

Pro-worker AI looks different. It supports decision-making. It enhances learning by respecting worker autonomy and privacy. It assumes workers are competent and curious, not that they need to be watched.

Where This Goes From Here

The research makes a strong case that healthcare and education are the two sectors where pro-worker AI could have the most impact and where public policy has the most leverage to shape how it develops.

Think about it if we designed AI in healthcare, to help doctors make better diagnoses, to give nurses more time for patient care, to help medical technicians do work that previously required specialists. We’d have healthier outcomes and more equitable access.

The same is true in education. AI could personalize learning. It could free teachers from grading to focus on students who are struggling. It could make expert instruction available to people who can’t afford it.

But none of that happens by accident. It happens because people choose to build it that way. It happens because institutions decide that worker wellbeing and skill development matter as much as cost reduction.

What You Should Do Now

If you’re reading this and you work in a field where automation is creeping in, here’s my advice, don’t wait to be displaced. Start paying attention to what’s the next problem? What expertise am I building that AI can’t replace?

The authors of this research are right about one thing: “the complementarity between machine capacity and human expertise” is where the real value lives. You are the human in that equation. Your job is to stay curious enough, skilled enough, and intentional enough to keep yourself in the collaboration, not out of the picture.

Learn how AI works. Not how to code it, not to say you can’t but your focus should be on how to think with it. Ask it questions. Use it as a coach. Get comfortable with it as a tool that makes you better, not a threat that replaces you.

And if you’re building AI in your organization, build it for augmentation first. Build it to make your workers better at what they do. Build it to free them up for harder problems. That’s not just ethically right, it’s better business. People who feel augmented stick around. They perform better. They innovate. People who feel replaced leave, or worse, they stay and resent you.

The future of work will be about choosing what kind of partnership we build between AI. Pro-worker AI says we can choose better. We can choose augmentation over replacement. We can choose growth over efficiency alone.

The question is, will we?