After AI... Are You Still Playing?

I think I may have ruined fantasy football for one of my colleagues this week.

I was co-teaching a class with Dr. Sam Hakim, who’s a huge sports fan. He was telling our students how excited he was about using Claude to help manage his dynasty fantasy football team. For those unfamiliar, dynasty leagues are a version of fantasy football where you keep most of your players from year to year, so there’s a lot more long-term strategy involved.

Sam was using AI to analyze player statistics, evaluate matchups, and hopefully gain an advantage over his competition. I asked him what made the best players in his league so good, and he explained that the guy who runs the league is incredibly committed. He knows the players, studies the statistics, and follows just about everything happening around the NFL. Sam simply doesn’t have that kind of time to devote to it, so using AI to help level the playing field made perfect sense.

But naturally, I had to push the idea a little further.

If Claude can analyze the statistics and recommend who to start, why couldn’t it just make that decision? Why couldn’t it evaluate trade offers, monitor the waiver wire, decide which players to draft, and manage the roster throughout the season?

I explained how, with the right tools and access, an AI agent could potentially manage the entire operation. Sam could put his dynasty team almost entirely on autopilot, checking in occasionally to see how things were going.

This raised a question I posed to the class: At what point is Sam no longer playing fantasy football?

Where do you draw the line?

There’s absolutely nothing wrong with Sam using AI to help him compete. I think it’s a fantastic use of the technology. He’s overcoming a limitation that has nothing to do with his intelligence or ability to make good decisions. He just doesn’t have the time to study hundreds of players, and now he has access to something that can do much of that work for him.

But there’s a difference between asking AI to help you play and asking AI to play for you. And I don’t necessarily think one is right and the other is wrong.

Maybe Sam enjoys making trades but hates researching statistics. Great, let AI handle the research. Maybe he loves drafting players but couldn’t care less about setting his weekly lineup. Automate it. Or maybe he just wants to win his league and doesn’t particularly care how he gets there.

The important part is that Sam gets to decide. He determines which parts of the experience he wants to participate in and which parts he’s comfortable handing over.

That’s agency. And I think it’s about to become one of the most important conversations we have about AI.

Because fantasy football might seem like a silly example, but the same scenario is beginning to play out across nearly every kind of white-collar work.

Think about what you actually do at your job. Not your title, but the work you perform throughout the day. Reading emails, analyzing spreadsheets, preparing reports, reviewing contracts, managing projects, and making decisions.

If you spend most of your day sitting in front of a computer, there’s a good chance AI can already perform meaningful portions of your work. Some capabilities are still unreliable and require substantial oversight, but the progress over the past couple of years has been remarkable.

For decades, we’ve used software to help us do our jobs. We enter information, navigate screens, interpret results, and make decisions. The software helps us along the way.

Now we’re entering a world where the software can increasingly do the work itself, and we’re the ones deciding whether, when, and how to intervene.

An accountant might use AI to reconcile transactions but personally review anything unusual. A marketing director might let AI develop campaign concepts but insist on making the final creative decisions. A business owner might be comfortable letting an AI agent handle customer inquiries and scheduling without asking permission every time.

Everyone draws their own boundaries based on the work, the risks involved, and how much they trust the technology.

But here’s where the fantasy football analogy starts to break down.

What happens when you’re not the one deciding where that boundary sits?

Agency works both ways.

For quite some time, my personal mission has been helping people reclaim agency in a world being reshaped by AI. I’ve spent a lot of time thinking about what that means, and the more capable these systems become, the more complicated the question gets.

When Sam uses Claude to eliminate hours of research, he’s gaining control over his time. When a business owner uses an AI agent to handle repetitive administrative work, they’re potentially doing the same thing.

But what happens when that business owner decides to automate work that an employee was previously responsible for?

From the owner’s perspective, they’re exercising agency. They’re deciding how to operate their business, manage costs, and compete. From the employee’s perspective, they might be losing that same agency, or potentially their job entirely.

Same technology. Same automation. Very different experiences.

I don’t think there’s an easy answer, and I certainly don’t think we can resolve it by declaring that AI should always keep a human in the loop. That sounds reassuring, but what does it actually mean?

If an AI can consistently make better decisions than a human, does that human really need to approve every recommendation? And if they become so accustomed to accepting those recommendations that they stop questioning them, are they meaningfully exercising judgment anymore?

Having a human involved in the process isn’t necessarily the same thing as preserving human agency.

That’s a distinction we’re going to have to wrestle with, both as individuals and organizations.

I don’t think we have as much time as people assume.

I feel compelled to make a prediction that might make some people uncomfortable.

I think we’re getting close to the point where AI agents can handle substantial portions of white-collar work with far less human intervention than most people currently imagine.

I’m not suggesting every office job is about to disappear. Real businesses are complicated, and there are legitimate challenges involving reliability, security, regulation, and accountability. But I don’t think we need to solve every one of those problems before the economics of white-collar work begin changing dramatically.

And I don’t think we’re talking about ten years.

My bet is that the next major advances in autonomous knowledge work will be measured in months, not decades. Double-digit months is where I’d put my money for significant changes in what organizations can realistically automate, even if widespread adoption takes longer.

I could be wrong about the timing. Predicting when technology will cross a particular threshold is notoriously difficult. But I’ve spent more than 3,000 hours over the past two years building with these systems, and what I’ve experienced firsthand makes it difficult to believe we’re another decade away from some pretty fundamental changes in how work gets done.

We’re rapidly approaching a point where the bigger question isn’t what AI can do, but what we’re going to let it do. And perhaps more importantly, who gets to make that decision.

So what are we actually trying to preserve?

In last week’s issue, I wrote about how software is beginning to adapt to businesses instead of forcing businesses to adapt to software. That’s one reversal.

This is another. For decades, we’ve used technology to help us perform work. Now we’re developing technology that can increasingly perform that work for us.

Personally, there are plenty of things I’d happily never do again. There are also things I genuinely enjoy doing, even when AI could probably do them faster or better. Writing is a good example. I use AI to help organize my thoughts and challenge my ideas, but I still care deeply about what I say and how I say it. That’s not something I want to hand over entirely.

Which brings me back to Sam and his fantasy football team.

If the whole point is to win, maybe putting the team on autopilot is exactly what he wants. But if the point is to enjoy playing, automating everything might defeat the purpose.

Work is obviously more complicated. Most of us don’t get to choose whether we participate, and the consequences are far greater than losing a fantasy football matchup. But the underlying question is similar.

What do we actually want to spend our time doing? Which decisions matter enough that we want to make them ourselves? And what are we comfortable handing over?

I’ve come to believe that agency isn’t about insisting on doing everything ourselves. It’s about having meaningful control over the decisions that shape our lives, including the decision to let something else do the work.

The challenge is making sure we don’t surrender that control simply because we’ve become capable of automating everything around it. And making sure the people affected by those decisions aren’t forgotten along the way.

I don’t have all the answers, but I think these are conversations we need to start having now, while we still have an opportunity to influence how this unfolds.

Because the technology isn’t waiting for us to figure it out.