A Conversation with Juliette Powell: Author of The AI Dilemma

I think I make a lot of people uncomfortable. I think I get brought in specifically because I’m an outsider. I’m a woman of color. I’m 6 feet tall. I do not toe the line, and I tell people exactly what I think. That could be incredibly scary and intimidating, especially in a corporate environment. So I am the perfect example of what I’m saying is necessary. Because there’s a reason why people bring me in. Within the organization there isn’t enough diversity, so they have to find it elsewhere. No matter how you end up doing it – whether you hire for it, or bring in people from the outside, or you nurture an internal and external ecosystem which is probably the healthiest of all – try it. See what happens if you haven’t done it before. If you have done it before, then hopefully you have the metrics to reinforce it as a positive practice.

One of my go-to examples is -- and this is a huge disclaimer: it’s being run by somebody that I care for very, very much, Astro Teller -- X, which is a division of Alphabet and is, in my opinion, incredibly well run. If you're not familiar with X, they're the moonshot factory. [They develop] so many projects that are big picture and are good for the world, not just good for the company. Because their timeline is much longer -- it's not about the quarter or the year -- success means something else than it would for their products and services divisions. I love their approach, and I think that they really have their heart and their mind in the right place. And again, this idea of bringing in just very, very divergent people and cross disciplinarians, too. People that know how to do more than one thing mesh well with others that know how to do more than one thing, and one of the key elements there, which I think we can all take a lesson from, is leave your ego at the door. You need to be able to play well with others. You're not the only expert in the room, and you have to be able to leave room for other expertise.

Another company that does it really well that I just discovered -- they're not even considered a technology company -- the Lincoln Center here in New York has been incorporating more and more AI, not just into their processes, but also in their shows. I was just there a couple of nights ago for the opening of the Malcolm X opera, and they had fellows from across New York, really a very, very different group of young people who were essentially working at nonprofits of various kinds, and they were being brought in to the Lincoln Center in this fellowship program for two years, where they get to mix and match and help each other's nonprofits, but at the same time they will be placed on boards of Lincoln Center resident organizations. They’re not just reviving the arts in New York, but also diversifying the arts in New York, and really starting at the top. I think it's one of the best examples I've seen in the city of really actively injecting all types of diversity in action simultaneously.

If I were to write the book right now, I would write the entire book about the calculus of intentional risk. This idea that different organizations of different sizes have a very different reputational risk. If you're deploying models that are coming from third parties, how much of your intellectual property is being leaked out? What happens if regulation comes down the pipeline and you get sued? What does your insurance say about any of this stuff? Is big tech really indemnifying you or not? How much does it actually cost to scale generative AI within your organization for a particular use case? If you are using AI, do you really have to be using AI? All of these things are part of a larger calculus of intentional risk. When companies look at their AI governance and they’re doing their risk benefit assessment, how long term is it? Are they also looking at the potential of refining the model? What happens if it keeps hallucinating and doesn't end up getting better and it ends up costing them far more than they anticipated? Are they calculating that? And on the other side, are they calculating what it's going to cost them if they don't take these other risks in the short, medium, and long term?

It’s not just data; it’s people. The data just describes actual human lives. For financial services, you might think of the data as money or information that surrounds money. But, beyond that, it's people's savings. It's people's lives. It's people's hopes and people's dreams. That's why your responsibility is much broader than whatever the regulators put on you. Because this is new technology. Nobody is an expert. So don't just play with it because you've got a new shiny toy to play with. Play with it in a responsible way that will help benefit the majority of people that support your organization.

I would recommend Joy Buolamwini’s book, Unmasking AI. I wrote about Joy in my book. She taught me a lot, and I've never even met her. She did her PhD at MIT, studying robotics. She literally had to wear a white mask on her face for her robot that she had created to recognize her. Her book came out on October 31st. I haven't read it yet, but her story is amazing.