Privacy and the Public Sector: Sapient Government Services’ Teresa Bozzelli on AI’s Government Trust Gap


Artificial Intelligence has rapidly moved from the comic books and movies to the stuff of real-live business applications, and the public sector is among the industries AI stands to affect most, particularly around data. But there’s a disconnect between the information we’re comfortable giving to private businesses and what we trust the government with. That gulf is resulting in a slower adoption of the technology in the public sector, and delaying a host of efficiency and cost improvements technologists believe would come along with it. Consulting caught up with Sapient Government Services’ Teresa Bozzelli recently to talk about this trust gap, how the government can overcome it, and how AI will eventually permeate the public sector right alongside private.

Consulting: What are you most excited about regarding AI’s entry into the public sector?

Bozzelli: One of the things I’m most excited about now is the fact that this administration is willing to push the envelope of what is innovative and progressive and disruptive in the private sector in our commercial world and having much more aggressive discussions around how do we bring that into the public sector and what might be the barriers around that.

I think this is the right time and artificial intelligence and data mining, those kinds of things I think are foremost in that discussion of where we can do that. I think the private sector is ahead of us in the usage of AI for ensuring better connections and experiences with a consumer to a retail organization, or better, faster services in detection of bad behavior like fraud. Now it’s time to bring that into the public sector.

Consulting:  What are some of the ways AI has already become prevalent in the public sector?

Bozzelli:  I do see AI as something that’s going to be able to improve mission delivery, something that will create a better customized experience for citizens, it will be able to create a better situational awareness and therefore better proactive intervention in negative situations, whether that’s from a disaster or whether that would be in our intelligence or national security perspective. I think those are key areas where AI has the opportunity to really change the result and lower the cost of government services.

Consulting: How big an impediment will budgetary constraints be on the adoption of AI?

Bozzelli:  I think in the public sector we’re using our budget for the wrong things, so we tend to spend more money operating old, historical capabilities. In our technology, we’re spending more like 80 percent on old stuff, and the modernization legislation they put forward last fall really highlights the need to think about shifting their investment to the future as opposed to maintaining the past. I don’t think that budget is necessarily what’s going to hold adoption back in the case of AI, but where we stumble in the public sector around AI is around privacy and data ownership versus data assurance and data usage. For example, in the retail space and marketing there are data sources out there that know exactly what you buy, exactly what you watch, exactly where you are, and that allows a retail organization to target their marketing messages to you based on historically your behavior and where you’re at.

You may get a text message with a coupon for a store right as you walk by. That’s not an accident, that data exists, but there’s a hesitancy to bring that into government because that perception of lack of privacy is something we consumers are comfortable with, but when we put our citizen hat on we don’t feel comfortable if the government has that same information even if it’s publicly available.

The perception of privacy in the government is in opposition of the expectation of privacy in the retail world. That I believe is what is going to restrict or govern the adoption of AI, and the same would be true with machine to machine communication as well. Sensor technology. Another example, the autonomous vehicle knows a lot about you that we as drivers don’t want that information made available to the government, but we’ll make it available to our insurance company.

Consulting:  How can the government overcome that trust gap?

Bozzelli:  The government is going to have to become very progressive about a data governing policy that just says this is what we will own versus what we will use, and that gets into gaining insight as a service. There are a lot of tools out there that operate on these large sources of data that either the government has or the government can discover information that’s relevant to their mission from publicly available sources without ever owning it, that’s an important part, to separate that data ownership versus data usage, and I think that’s a policy that’s going to have to evolve over the next 2 to 3 years.