Anita asked first about Henri’s experience with corporations. Henri focused on how companies engaged particularly during COVID. How we measure success is of course a big challenge, but this went relatively well. A lot was about State and corporate relationships around housing and transportation particularly. The housing challenge during the pandemic in NJ was about finding people places to go for quarantine when they tested positive. The State had limited options. NJ set up partnerships with AirBNB and hotels to create places where people could go. The State was paying them, and they met a massive need. In transportation, there was a focus on healthcare professionals who were in short supply. NJ partnered with American and Delta airlines to fly in medical staff from other states. Henri was able to work between the healthcare sector, airlines, and hotels to source healthcare personnel.

We then turned to AI designed for the private sector for use in government. Henri talked about a whole range of ways in which corporations work with government. Engaging with private sector in suppliers, vendors, etc., is part of this. Another is that they are filling a need in a moment such as COVID in a philanthropic sense. When it comes to AI, Henri talked about a dynamic of hesitancy in the public sector. The biggest risk is data privacy. There has to be a feedback mechanism in AI for it to be AI. There are great privacy issues on sensitive applications like health and tax, for example. The vendors are becoming more sensitive and are building in controls. Another mitigation approach is to create sandbox environments in which governments can work directly with the models. Anita worries also about bias and about doubling down on the established systems. The job of government isn’t necessarily to be creative, though. It’s to make the day-to-day experience of being a citizen better. If we’re just thinking about AI as a tool, then it can make government better. Can we use AI to make things simpler and clearer?

Henri and I brainstormed on how firms like AWS and OpenAI could work over the long run with governments on models and other AI-enabling technologies. We have to think through how this could work well. There’s a genuine interest in this on both sides, but we are at the beginning of this path. Government has a huge role on the application and evaluation of models for their use cases.

Anita asked Henri about experimentation across the public and private divide. One unique but replicated example is during emergencies, where you see bureaucracy fall back. That’s when you see a lot of innovation, such as remote work during the pandemic. This wouldn’t be happening without that catalyst. That’s one area of opportunity. Governments tend to be risk-averse. The consequences of changing a 911 system are very high. The risk seems to overwhelm the benefit. If a company can offload risk from government through experimentation, that can facilitate innovation. There’s a triangle where firms engage with a third party – perhaps a non-profit in the community – and then the government outsources problems there and allows experimentation to take place. There are also agencies of government with mandates for innovation, like Darpa and CIOs.

The Oakland example that Henri worked on emphasized homelessness, illegal dumping and illegal weapons, and the ways in which it was hard to implement even in partnership with local companies and government. There was a lot of muscle-building in this project that will pay off over the long run. The local government learned a lot about how to source citizen input and democratize access to government. One way to look at a project is on outcomes that are prespecified, but another is to consider culture and the way organizations think. How do you transfer cultures of innovation into government? Henri and the team pushed people beyond their comfort zones in learning how to work together. Experimentation in and of itself is a great outcome even if the measures on homelessness, illegal dumping and illegal weapons metrics didn’t change in the short run.

Henri wants to be ambitious but he’s also cautious. One thing he’s excited about is working deep within companies with people who want to work in a thoughtful way in good faith. AWS, for example, has provided a lot of pro bono support out of real commitment to the public service. It’s genuine. Being humble and needs-driven is a great aspiration.

Henri ended by reflecting on knowledge-sharing as a critical activity in any partnership. It should be about 20%. That becomes the foundation for real innovation.