I use AI tools a lot. Writing tools, productivity apps, coding helpers. I use them almost every day. So when I started watching Model Employee and listening to Timnit Gebru’s podcast, I thought I would feel fine about it. I thought I already knew enough about AI to not be surprised. Instead, I finished both feeling uncomfortable and a little embarrassed about how little I had actually thought about these issues before.

What Surprised Me

The biggest surprise in Model Employee was how detailed the surveillance is. I knew companies track workers in some way. But I did not know it goes this far. AI systems can monitor keystrokes, mouse movements, facial expressions, and even the tone of your voice. All in real time. All the time. It is not a manager who checks in sometimes and then goes back to their office. It is a system that never stops, never gets tired, and never looks away. And at the end of the day, it gives you a score. That score can affect your pay, your hours, or whether you keep your job at all.

What made this worse was thinking about the kind of jobs where this happens. These are not high-paying tech jobs. These are warehouse workers, delivery drivers, customer service workers. People who already have less power in the workplace. The AI does not make things more fair for them. It makes the imbalance bigger.

The second thing that surprised me was how hidden all of this is from people like me. I use AI tools and I think about what they can do for me. I never really thought about what they do to someone else. Timnit Gebru made a strong point in her podcast. The excitement and the media coverage around AI almost never includes the people whose jobs, routines, and dignity are changed to make these tools possible. There is a lot of talk about what AI can create. There is very little talk about what it costs and who pays that cost. I was one of the people who got caught up in the excitement without asking those questions.

The third thing that stayed with me was how the workers in Model Employee had just accepted it. Many of them had changed their behavior to perform better for the algorithm. They were not fighting it. They were adapting to it. That was not surprising in a shocking way. It was surprising in a quiet, uncomfortable way. It made me think about how quickly people normalize things when they feel like they have no choice.

The Connection to the Course

This connected directly to what we discussed in the course about AI ethics and power imbalances. We talked about how AI systems are not neutral. They are built by people, for a purpose, and they reflect the interests of the people who design and fund them. Model Employee is a very clear example of that idea in practice.

The monitoring tools in the video are presented as objective productivity measures. But they are not objective. Someone decided what counts as productive. Someone set the thresholds. Someone chose what to measure and what to ignore. Those are all decisions made by people with specific interests. And the workers being measured had no say in any of it. That is exactly the kind of power imbalance we analyzed in class, just applied to a workplace setting.

Gebru’s podcast connected to another thing we covered, which is how language and framing shape the way we think about technology. She talks about how AI is often presented as inevitable, as something that is just happening and that we have to adapt to. But that framing is a choice. It is a way of closing down debate before it starts. Our course gave me tools to recognize that move, and hearing Gebru make the same argument helped it click in a new way.

What I Will Do Differently

This was the hardest part to think about honestly. I use AI tools heavily and I have mostly evaluated them on one question: does this work well for me? Does it save me time? Is the output good? I did not ask who built it, what their practices are, or what the tool might be doing in other contexts.

After watching Model Employee and listening to Gebru, I want to change that. Not in a dramatic way, but in a practical one. Before I start using a new AI tool, I want to spend a few minutes asking some basic questions. Who made this? What do I know about how this company treats its workers? Is this same technology being used elsewhere to monitor or control people? What data does the tool collect and what happens to it?

These are not questions that will always have clear answers. But asking them is already different from what I was doing before. I think part of being a critical user of AI is accepting that my convenience is not the only thing that matters. The fact that a tool is useful to me does not mean it is good overall.

What the Course Should Cover More

One thing I think is missing from the course is what I would call the business side of AI. We talk a lot about the ethics of specific systems, which is important. But I think we also need to talk about the bigger picture of who profits from AI and how that shapes everything else.

Gebru touches on this in the podcast. The companies building and selling AI are among the most powerful and least accountable organizations in the world right now. The decisions they make about what to build, who to sell it to, and how to deploy it are shaped by profit, not by ethics. Understanding that does not mean being against all AI. But it does mean being realistic about why certain problems get AI solutions and others do not. Warehouse workers get surveillance tools. Executives get productivity assistants. That is not a coincidence.

I think adding a unit on the political economy of AI, looking at corporate structure, investment flows, lobbying, and market incentives, would make the course much stronger. It would give students a more complete picture of why AI develops the way it does, not just how to evaluate it once it is already here.

This reflection is based on Timnit Gebru’s podcast “Don’t Fall for the AI Hype” and the short documentary “Model Employee.”