Michael, thank you so much for that speech. I feel like I appreciate the choice to talk about inequality and to just dive into AI. So, I wonder if we could just start there. You laid out a lot, but when you think about the lens of people living on low and moderate income, is there any piece of all that you laid out that you're especially bullish about? Let's start with the upside.
Yeah, Tim, first of all, let me just say it's a great pleasure to be back together with you here. And I remember 25 years ago when you and Peter launched D2D and what became Commonwealth and the work that you've been doing over the last quarter century now is really been impactful. So great to be back here with you. Let me say, I think that if properly deployed, artificial intelligence can help low-income people in lots of different ways. So we'll talk, I think, in a little bit about this in the finance area, but more generally thinking about as workers, let's say you're a worker with low English language proficiency. AI can help you make that leap to using English much more quickly than if you had to take the full time to learn English as a second or third or fourth language. So that's an example. AI can help workers become much more productive and thus increase the possibility or probability that they'll get a wage boost from that productivity. So for example, there's been lots of work on workers in call centers and it's true that AI might diminish the number of call center workers, but it's also the case that the call center workers who remain doing that will be much more productive and be able to earn higher wages because of that productivity. So these are the kinds of things you can go not only job by job but taking the job apart and looking at tasks within jobs and ask how is AI going to make me more productive as a worker and therefore enable me to advance in my career in ways that if I had lower levels of education might be very difficult.
Those all resonate. Let's pick up on something from your opening remarks which I was really struck by. Because sometimes you have an anecdotal experience and then someone delivers data that really makes the point. And what I heard you say is that people who have less education and probably less income are using some of these frontier generative AI models less. And I think you even made the point that they're seeing less effect in the way that they're using them. And that leads me to this question which is we see in our work that there is quite understandable mistrust of institutions, of technologies, of systems, and that can be a factor that holds people back from trying things. And so I'm curious for your thoughts on that and what you think we can do to kind of build that trust or earn the trust so we see everybody, at least getting the same chance to benefit.
Yeah, it's interesting. I do think AI has the potential to increase trust appropriately in the financial system. There are lots of careful words in that. It has the opportunity to increase that trust appropriately. So what do I mean by that? AI is a tool and like other tools, it can be used to increase trust in inappropriate ways for you to trust a scammer, for example. AI can be used for fraud. We're seeing that already. Or cyber attacks. So you might inappropriately increase trust. But I'm talking about appropriately increasing trust in institutions that should be trusted, in processes that should be trusted. AI might help overcome some of those trust barriers. So there is some sense in which AI can connect with and respond to individuals in a tailored individualized way so that the person reaching out to the AI for help is being responded to as an individual human being, and in the same way that a one-on-one conversation with another human might help build trust, having that kind of tailored response that treats each individual for their own individual selves might also be a way to build that trust. Or I mentioned before, language can be a barrier to trust and AI can bridge that quite readily. Well, these are just a couple examples. I know Tim, you've spent a lot of time thinking about trust in the financial system. So how do you think about that same set of issues?
Yeah. Well, I can share a little bit of research we had a chance to do. This was really looking at banks and chatbots, but I think the lessons translate and some of what we saw is what you already highlighted and might be common sense. People want to be sure that their data is safe. Who among us doesn't want to be sure our data is safe? We know there are bad actors out there. So that made a lot of sense. Another thing we saw that people tend to trust technology if they trust the provider of that technology. So in this case that was banks and saw a very high correlation between how people felt about their bank and whether or not they trusted the technology. I think that finding probably translates to whether it's the technology firms or other financial service firms delivering the technology. But some of the things that were maybe not immediately obvious that I think are really important: people had a very high sense of whether or not the technology was useful and relevant to them and the experience. And again, how many of us can relate to this, especially if you think about an old school chatbot, you engage and you just get something back that's just not all that helpful. You know, oh, I could have done that myself or I can sense I'm in some kind of decision tree and I've been pushed to an answer. People are very perceptive about that even if they don't have a lot of money and they go, wow, this is not for me. So I think that's important to your point about personalization and tailoring; people needed to be useful to them. The other thing that we learned, which again maybe surprising maybe not, is that people need to feel that they're not being judged. And I think most of us carry some kind of complicated feelings around our finances. If we have less and we come from less money, we probably have even stronger feelings. And so I'm personally bullish that generative AI has the conditions to create that feeling that you're not being judged. And then the last thing which in some ways was most surprising is people told us they want the experience of using the technology to leave them feeling more competent, feeling better and more confident about their own abilities. And I'm not sure we know exactly how to solve for that, but I think the generative AI tool set gives us more of an ability to actually help people walk away and go, I learned something. I understand something. I feel that I've had a real conversation to your point, Michael. And that might be something we overlook, but I would love to see how we keep this part of the conversation because it does worry me that a whole huge swath of our society won't engage with these tools for very understandable reasons and then the advantages will just compound to those of us who are able to.
Yeah, that's exactly right. I mean we need to see broad adoption in order for it to potentially have that positive effect that both of us have been describing. And I very much resonated with your point about individual capability. People want to feel empowered by the technology they use to become smarter, better, faster, stronger, whatever your definition is. And so I do think that's a potential for the tool again if appropriately designed and used.
And I appreciated in your remarks you talked about the need for greater curiosity and some other just briefly, I think a lot of us in this room who care about these issues and your long history are aware of many resources that have gone into financial education, financial capability building. And I'm sort of curious how that needs to evolve as the tools and the challenges evolve. Do you have any thoughts on that?
Yeah. Well, first let me say just a word about curiosity. When young people ask me about their careers and what they should study, I spent a lot of time in academia and mentored a lot of young people, the first thing I say to them is the most important skill you have is a sense of curiosity because that's something you can carry with you your whole life. And if you have that, you're going to learn new skills all along the way. One of the amazing things about being here at the Federal Reserve, I know when I come to work in the morning that by the end of the day, I'm going to learn something new. That's pretty amazing. So that's the curiosity. Now, I got distracted with that. Your second question is about...
Is there something that we want to be thinking about to evolve financial capability, financial education training so that we're actually setting people up?
So I think financial education has moved already a lot over the last 30 years. It used to be the case that people talked about financial literacy and the model in people's head was you bring people into a classroom and you teach them a bunch of things and then they go out and use it in the world. And over time it became apparent that that was not the smartest way of thinking about financial education. And the field moved to effectively financial capability and empowering individuals using a moment of decision, using behavioral insights, and using some education to change outcomes. And I think AI, generative AI will accelerate that movement in that direction because now you can use that moment of contact and that behavioral insight in a way that is even more tailored to the particular circumstances of the individual being affected. And that makes it more likely that you'll get a better outcome. We originally started talking about financial literacy many decades ago because we wanted better outcomes for people and now you can get to those outcomes in a more potentially in a more direct way.
I appreciate that and I guess the one additional thought I'd just add is that I think again in this long history out of an abundance of earnestness often we have wanted to make sure people are equipped with all the right facts and knowledge and we've learned that's not enough. And I think one frontier we want to be thinking about is a little bit what we talked about just a minute ago. We need people to both have the basis to be confident and to build that confidence in their own ability to navigate. And that's really tricky. I'm not sure exactly we know how to do it, but I think putting that on the agenda. You and I talked a little bit about this. We do a lot of work trying to support people who want to become investors for the first time. And through that, we learned that one thing they'll say if you sit long enough is, 'I know what to do,' and I actually have a little bit of money and I know what tool to access, but I still don't feel like I belong in that world. And those kinds of subjective realities are invisible barriers and I think we want to figure out how we make progress on those. If I could shift, we're thinking at least in our shop, and I think you do too, a lot about the financial lives of people who don't earn a lot of money. And one thing that just comes up over and over again is kind of having a little bit of slack, just making it through the week or the month. And do you have any thoughts about what role generative AI can play in increasing that slack for folks?
It's a great question. I wrote a book about this topic back in 2012 called No Slack, the financial lives of low-income Americans. And the reason I wrote that book is because I had seen the lived experiences of low-income people who were really smart and were hardworking and were just trying to juggle in an impossible way their income and expenses. And that lack of slack made it much more likely that a small problem would lead to a big one. And more broadly, Sandel Mulan and Eldar Shafir wrote a book a few years back called Scarcity where they explore that concept not only for people who don't have any money and therefore don't have financial slack but people who have scarcity of other kinds. So you can have a scarcity in time and therefore not be able to do the things that you need to do. Or you can have a scarcity of focus. There's only so much that we can focus on at any given time. And you may need to focus on one set of issues and therefore be blindsided by a problem that comes up. So AI could help with those kinds of scarcities. So on the financial side, AI could help individuals better plan, better anticipate problems in the future, and therefore be better able to ensure against those difficulties whether through savings or through insurance or through credit to smooth through those difficulties. So AI might be sort of an uber budgeting tool for low-income people. It could turn out that AI changes the financial slack people have directly by giving them an outlet to be more productive and therefore to earn higher income and therefore to be less stressed in the lives they live. And AI might help also in the scarcity of time and of focus. So if you have scarce time and focus and you're really focused appropriately on doing your job, for example, as an auto mechanic, you don't have much time to think about these other things. Well, AI can be doing that work for you while you're doing your day job, it's doing its job of paying attention to this financial issue you're worried about and spending time solving that issue. So it really can help with all three kinds of scarcities we've been talking about. And I know you've also been working on this for 25 years. So how do you think about it?
I love all three of those and one additional I'd nominate is I think there's something about perspective that can be really hard for all of us. And whether it's scarcity or your own research or just simple intuition, when we're under stress, our worlds get smaller and closer and more immediate. And at least in theory, this technology offers a way in a non-judgmental private space to zoom out and think about the whole arc of your financial life. And that sounds a little abstract, so I'm not sure we know how to operationalize it yet. But even a question like adding to your income and being able to instantly model what that will mean for your future, or I was having this conversation with somebody recently who was saying, you know, I think if you want to remember there are families that have never had certain opportunities and if you say to parents, if you do this and you do this and you do this over this period of time, your child could have enough for a down payment by the time they're 25. Like think how motivating and how horizon opening that could be. And now we have the tools to actually back that up with a concrete set of actions and steps and to model it out and to place it in context and pressure test it if these things happen. And we got to figure out how to make that work for people at a sort of user experience level. But it is exciting to think about the chance to help people zoom out and think in that way, I would argue. Yeah.
Can we switch gears? You know, we're here at the Fed and you all have such an important role in thinking about the welfare of the entire economy. Let's talk just a little bit about crisis and shocks. In my own very limited experience as a person and a student, the irony of shocks is that they're going to come, right? But you don't know when. You don't know exactly what. You don't know how this one will be a little bit different than the last one, but when we think about this lens of people that don't have a lot of money, what keeps you up at night? What are the things that you think we should be preparing for and how does generative AI in particular potentially change that? And I know you spoke to some of it in your remarks, but I just wanted to give you another shot at it if you're game.
Well, one of the things when we're looking at the financial system as a whole, one of the things that we do is we try to focus on the vulnerabilities in the financial system because we don't know quite where our shock is going to come from. And I think the same is true for households. If households have enough resilience to withstand the kind of shocks that are in the normal range of shocks that people experience, your car breaks down, you need new tires, things like that. You don't know which of these shocks is going to hit you, but you know some of them will over some period of time. So if we focus on resiliency and reducing the vulnerability of low and moderate income households, that's a way of addressing the fact that any individual household doesn't know what shocks are going to hit them.
I appreciate that. That sounds in some abstract way like insurance, right? Like to look at the risks. Let's open it up to questions in a minute or two. So I just want to plant that seed. But I want to go back to something that you said in your opening remarks that I think is really really important and maybe easily overlooked. You said something about our shared values, right? And also in the alignment question, that strikes me as really foundational. In any pluralistic society we're never going to have just one set of values. But in this conversation about financial inclusion, what strikes me is that in some ways financial inclusion is an input. It's a status. And what we really want to get to is what's the shared outcome that we want. And I'll put my cards on the table. I want to live in an economy, in a country where most of us have enough, where we have stability, where we can show up fully in our households, in our workplaces, in our communities, in our civic life. Do you have any thoughts? And I realize this may be at the edge of what's inside our mandate, but how do we work towards a common set of values of what we want for households and for our economy?
