Tell us about yourself. What are you working on right now?

I’m a staff product operations manager at GitHub. My role is to help support the AI and productivity group. Productivity is kind of anything that would be considered a frontend or key interaction within GitHub. 

GitHub is very developer-focused right now, but longer term, I wonder about how everybody works together. That means not just developers and designers and product managers, but the legal team and privacy and all those types of things. Product operations is an interesting role. It's newer, but it's really about team dynamics. I PM the PM experience, and I think about not only how we make decisions, but also how we have discussions. How do we provide space for that?

How has your thinking about products and product management changed over the years? 

I was always fascinated by trying to understand problems and how we solve them. [What’s] changed over time is [that] once I thought I could think my way through everything and be perfect. But that is not the case when it comes to reality, the complexity of the world, the uncertainty of the world. And so I've started to embrace more of this idea that we don't actually know anything.

It's much less about “I know for sure what is going to be here” and it's more about “How do we discover that?

I'd say every place that I join, every project I end up working on, it's much less about “I know for sure what is going to be here” and it's more about “How do we discover that?” So at GitHub I've been interviewing every product manager in my immediate group—and I've talked to 24 people so far—just to ask, “What is success here?”

Maybe that's my mission overall: How do we have better conversations? How do we agree to things in a way that are helpful? The future that I want to try to bring about is one where we can actually make decisions, and be okay with not knowing as well. 

What was your first encounter with AI?

I was always really fascinated by bulletin board services and bots and stuff like that. So even on IRC way back in the day, there were these agents that could exist within a human space and they would do things. They were very simplistic. Then in college, I was a TA for someone that did a lot around network theory and the idea of resilient networks. That was the first time that I really started to think about the way that things work together in that way. Now, AI, of course, could be something like a spell-checker, a chatbot, even image analysis—all these different things would be classified as artificial intelligence.

When chess was basically solved by a computer, we started thinking that there was, like, a possibility for centaur human beings [to be the ultimate chess champions]. That’s not true, actually—no one can ever beat any chess computer at this point. But there are more people playing chess than ever. People enjoy it. It has also changed the way we play these games, as found after the open sourcing of models with Go. In these cases, I think the idea of being empathetic, compassionate around human beings actually is what we should be leveraging humans for.

How do you use AI in your work?

The main thing is when I need to have provocation: it helps me with creativity [and] reflection. I use the card deck Oblique Strategies a lot. And [the experience of using] Oblique Strategies is something that ChatGPT can actually create. Most of [what it generates is] garbage, but there's [always] a few in there that end up making me think of different ideas.

If I wanted to summarize this to one sentence, what would the machine think it was?

Summarization is another [way I use AI]. If a product manager [feeds AI] their entire document and it gets summarized down to something and it's totally inane, [that means] you probably don't have a good document. There are these good double checks: if I wanted to summarize this to one sentence, what would the machine think it was? It’s very eye opening. But what I don’t use it for is, I don’t go to ChatGPT and say, “Write me a spec for that thing.” Because that’s usually boring.

You also write fiction. (Allma's magazine, Roadmap, published your short story “Return to the apocalypse” earlier this year.) The traditional idea of writers with day jobs is that their writing and their job are very separate, but for you it seems more integrated. How does your work inform your fiction?

One of the first stories that I published was “Mechanical Turks All the Way Down,” which was about someone being asked to do something, but it ends up being actually a machine that's asking them to do something. So this idea of layers, where are you in this hierarchy of machine versus human, and what does it mean to be human inside [of it]? I think the reason why fiction is so integrated with what I think about [at work] is because I am constantly struggling with that identity issue.

My story, “Return to the Apocalypse,” is really about how much we put our identity into work. And literally, in the life-and-death end of the world, this person still had so much identity caught up in [their job].

How does the speculative, imaginative, narrative part of your brain inform your work? Why is that skill useful?

Everything that a product manager does is fictional until it comes to be. I've had people disagree with me about that! But I think the speculative aspect is that we can imagine futures that are both worse and better than they are today and that allows us to make decisions about what we want.

Everything that a product manager does is fictional until it comes to be.

Is this far-flung future in alignment with [our] values? And a lot of people don't even have discussions about what their values are. That's what I end up using [my interest in the speculative] for, [because] it's an easier way for us to have a conversation about values. 

There was a group inside of Google called “Moral Imaginations.” And for a machine learning [team], they would have short story writers create a fictional scenario, an extreme version of whatever technology we were creating. And then we would role-play: one group would represent a hypothetical, Google-like company, one would represent the local government, one would represent journalists. When you take on role-play you are allowed to be something else. And a lot of the time, especially at Google or Microsoft, these large companies, people feel like they have to be a particular way and so they just sleepwalk through decision making. But if you give people the option to be something else, they can think of things differently.

Have generative AI and the tools that are available to everyone today changed product management at all? How do you think it will change?

It's going to affect everything at some point, but I don’t think it’s actually impacted product management as much. People like the idea that there’s an AI product manager, but I disagree. I think there are some new rules and heuristics we need to develop, but then it will all just be product management again.

Do you have any tips or strategies you can share from how you are using AI now? What advice would you give to other product managers?

Don’t believe the hype. There’s a lot of issues with productionizing these systems. Air Canada got sued because they had a generative AI chatbot and it made up a policy and then they had to adhere to it. There’s a lot of experimentation going on right now, and that’s great. But [turning] these things [into products] is going to take a lot of work. We should get down to [the question]: “Is this solving a human problem or not?”