AI is a foundational technology. By now we all know that infrastructure is not just bridges and railyards; it’s family leave and health care and much more. The point is that infrastructure is the foundation for other things. We can't build our society without it. When we think about AI, think about it as building the infrastructure to build other things.
Then think of it as public AI—not just private. This was the mistake we made with social media. We really need a public AI option, or options, to ensure the creation of AI systems that are trustworthy, open, and sustainable. This technology will be the foundation for our digital life and increasingly our physical lives—the line is blurring already. We need to think about its development systematically in terms of normal technology, as Arvind Narayanan and Sayash Kappor have written, rather than giving in to all the hysteria around it.
What does public mean? It’s not just the public sector but also the public interest, and that includes bringing in the private sector. I heard many of you use the term “public” in this way; indeed, there was remarkable consensus on this.
Public AI, quite possibly developed in partnership with private companies, can make markets more competitive—make it possible for smaller enterprises to compete with larger enterprises. And for people who are not advantaged in our traditional economy, they can be better included; imagine integrating AI with existing infrastructures like libraries. Let’s also not forget about non-tech legislation like public records laws, which are not meant for AI, but they can help here.
Partners, partners, partners. The need for partners kept coming up during the past two days. Partners for what exactly? It’s easy to create partnerships; it’s not easy to use them effectively. You need to know who you are partnering with and why, and you need active partnerships. Organizing, prodding, de-conflicting, making sure you are producing something for the same specific reasons. When can we start building AI tools together? That is a question we all need to ask. We’re not ready, in many instances, to partner just yet.
Maybe this is all a race that will be defined by the tortoise rather than the hare. This is an area where you can benefit from learning from others. AI is a marathon, and for that marathon, you need to build confidence and plan for the long term.
Another point on AI: Play! Just use it. Have fun with and try AI for everything. A lot of people playing with the tech can be just as important as building the infrastructure. I can tell you, I will be playing with it more this summer.
On AI governance, I heard a desire for minimum viable governance—start with something and develop it. There is an entire body of administrative law on “rolling rule regulation.” It started over two decades ago—applied to labor standards and then environmental law and utilities. It’s a dynamic and iterative approach to regulation that evolves over time with continual monitoring and adjustment. So we don’t need to invent this model for AI. Just be willing to say, “I don’t know.” AI is changing as fast as the COVID virus. So the AI governance goal is not perfection but continuing to iterate and build trust.
Workforce and job displacement: This is a huge issue, and we have seen it play out—always with the same structure. There is a binary of “jobs will be replaced” or, on the other end, “we just need to adapt;” it is one or the other. We can’t look away from this issue even if we don’t know what to do. We need to assume the worst-case scenario, even if only a third of what’s predicted happens. It makes me think of Oren Cass’s book The Once and Future Worker. I don’t always agree with his politics, but he discusses how job loss leads to depression and drug addiction, which goes up for men. We can’t afford to not be ready.
A more positive dimension: There will be a real premium on human work—what AI can’t do. I spoke to a representative from Maine, and he said he has 10,000 people on a waiting list for behavioral health care—so there’s an enormous need in the care economy. Coaching, which may have been just for teens before, is now needed everywhere to enhance personal and professional human performance. From career to nutrition to education to habit-breaking and making—we are all going to have coaches. It’s a huge area for the creation of new jobs, jobs that will be better and more effective with AI built in. As we think of job displacement, we need to think about how AI can create and improve jobs.
AI and innovation. How do we create new jobs by creating new industries? The New Jersey AI Hub is about seeding innovation across the state, so how does it create new jobs? The innovation economy is about connecting—community colleges, business, civic groups, local government—and AI can help with building that new ecosystem. That requires focusing on AI that finds and strengthens connection, a human quality that is more associated with EQ than IQ.
I was struck by how politics was not mentioned at this convening, and that was such a blessing. There was a lot of optimism—not a focus on red versus blue states. I heard everyone talking about the same problems: needing to rebuild trust—not talking about a big or small government but a government that works well. That is worth bottling; I want to bottle the spirit of this conference. AI leaders can carry forward this spirit; the whole country needs that.
AI is a transformative technology. It doesn’t just transform how government does things: better, faster, cheaper. It can transform what government does and, even more importantly, what government in a democracy is. In a federalist system like the U.S., we have laboratories of democracy in the states.
Different states are experimenting to enable co-governance, approaches to self-government that “break down the boundaries between people inside and outside government, allowing community residents and elected officials to work together to design policy and share decision-making power.”
AI is a great equalizer here and can help us start co-creating and co-govern in many ways. New America has done a lot of co-governance work—not with AI, but with citizen assemblies and participatory budgeting. This is about organizing horizontally rather than vertically. That requires more and greater trust, transparency, and public AI. If we can build that trust and make sure AI works for everyone, we can really change what it means to govern ourselves with the help of technology—but putting humans first.
To learn more about what took place at “Shaping the Future of AI: A National Gathering for State AI Leaders,” be sure to read Princeton University’s article on the event.
Image: Anne-Marie Slaughter - CEO, New America