Dan Goldstein: Microsoft Research

For instance, you don’t know if you spend an extra five bucks a day every day if that will leave you poorer in the future or if it will make no difference. But computers are very good at figuring that kind of thing out.

Similarly, it’s hard to imagine how the future might look. We tend to neglect or disbelieve that we’re going to get old. But when a computer takes a photo of you and ages it, it makes it quite easy to look at it and say, “Oh, yep, that’s me. I really do need to think about the fact that there’s a future me out there, and future me is going to inherit all of the decisions I make today.”

It’s equal parts helpful and a little scary to think that $4 cup of Starbucks you buy every day could be affecting whether or not you have a comfortable retirement.

Right. And part of what’s scary is that you don’t really know if things like that do matter or don’t matter. You don’t want to worry too little and you don’t want to worry too much. You want to worry the right amount. That’s what technology is great for. It’s great for doing projections and letting you know this aspect of your life that you can change really has a consequential effect on the future you whereas this thing, that you may be spending time optimizing, doesn’t really matter either way: you might as well forget about it and do what makes you happy.

How many people decide to save more after using your technology?

We know that in our experiments people said that they would save about 50 percent more for retirement after being shown an image of themselves aging compared to a group that was not shown an image of their future selves. So we think that’s promising and what’s nice about it is that every employer typically has a picture of every employee and it is in the employer’s office where the employee is making this decision of how much to save for the future. We also know that most people never change that decision they make in that first week of work when they decide how much of their paycheck they want to contribute to their retirement fund.

I’m guilty of that.

Like 90 to 100 percent of people never change what they set up in the first week. But if you can show them that aged picture in that moment when they are making the decision, that causes them to save 50 percent more, and they stay in that job for 10 to 20 years, that’s going to have a huge impact on the amount of money they have saved by the time they retire.

We’ve been working to try and do field studies—we don’t have anything completed yet—with companies. But that’s one of our goals, to get this in front of people when they are making that decision because one little nudge at the right moment could boost their savings.

You mentioned briefly that you see other applications for this technology. What are some of those other research uses for VR technology?

We’ve shown people their future face, and now we’re showing people their future body, if you will. This is also work that I’m undertaking with Hal. A lot of medical spending in the U.S. is due to behaviors that people have as opposed to inherited conditions. So things like smoking, being overweight, not exercising are behaviors that can lead to health problems that can be very expensive to our society. Since these are behavioral problems, there might be behavioral solutions to
help people.

This future body experiment, which we are running right now, involves people who are trying to lose weight. What we have created is a system by which they can step on their scale in the morning—their scale is tracking their weight day by day—and when they start to gain weight, they’ll get an email that shows a rendering of how their body will look if they continue to gain weight. So they’ll see a heavier version of him or herself.

At the same time, if they are overweight and doing a good job losing weight, they will be emailed a computer rendering of how their body will look if they continue to lose weight. It’s using positive reinforcement and negative reinforcement by showing you that your body is heading in a certain direction but you’re getting that feedback immediately—within a few days of the weight gain—as opposed to the situation we face in everyday life: we gain weight very slowly and then we look down a year later and realize we’ve gained 20 to 30 pounds. This lets you glimpse into the future by sending you an email from the future of how you might look if you continue down the road that you are heading. The basis of the experiment that we are running is to see whether these kinds of images from the future—if you will—will motivate people to lose weight.

Seems similar to your finance research in that it can be helpful and a little bit scary about how much they may have gained in a specific time and what that means for the future.

We’re trying to be careful to not make it too sensitive to irrelevant changes in weights, but respond to what we think are trends that they need to be concerned about.

Do you think this type of VR technology has any other business applications?

I think the health thing is definitely a business application. Companies are the ones paying for the bad health outcomes often caused by behaviors that people have. Getting people healthier can lower the cost of providing insurance to people, which is a big part of the economy.

What would you say is one of the most beneficial uses of VR today?

I’d like to think that the financial aspect is the most beneficial. Humans aren’t used to thinking about compound interest, correlations between asset classes and things like that, whereas we can simulate out pretty well using computers financial futures and represent this back to them in a way that is easy for them to understand. I think that right now this is where we are seeing the biggest benefits, but I also think we are going to be thrilled and delighted by the kinds of things VR researchers are dreaming up and trying out to address a myriad of problems in life.

I’ve heard of people using VR and AR headsets to help get rid of the fear of public speaking. You put on this headset that makes you feel like you are in a lecture hall with 300 people and you can practice your speech and get a feeling for what it’s like to address a room with a large audience. Any kind of situation which is expensive to reconstruct in reality might be inexpensive to construct in virtual reality.

What are your hopes for the future of VR?

My focus has always been on decision making, so I really hope that for the future of VR that it can promote better decision making by just making it easier on our brains—showing us the outcomes that could happen as opposed to us trying to imagine these outcomes ourselves since our imagination may be inaccurate or uninformed.

Do you think this is something that could lead people down a rabbit hole of VR? That they will turn to technology for an opinion on everything?

I hadn’t really thought about it as a potential problem. We’re trying to take the most consequential decisions people face and tackle those first. That’s why we’re starting with things like your health and your finances. In the future we see people working with this technology to make big decisions. But I can see smaller decisions being aided with virtual reality in a way that wouldn’t necessarily constitute a problem. For instance, if you could “try on” clothes using VR, see how they look and fit, and by doing it in the VR way in five minutes saves you a three hour trip to the mall
and buying something and trying it on and returning it, I think that’s a good use of people’s time. If you’re making good decisions in less time then its freeing up time to do other things.

What does a normal day as a Principal Researcher at Microsoft Research look like?

When I was a professor, I was teaching, I was doing university administration, and I was doing research. Now the first two are gone and I’m just doing research all the time, which is a dream for me.

But what that means in exact terms is that on a typical day I’m meeting with collaborators, we’re designing experiments and running experiments, working out the logistics with partners if its a field experiment. Once the data are in I’m analyzing the data, doing statistical analysis on it, graphing it, figuring out the best visualization, writing it up in the form of academic papers, and submitting those papers for publication.

Can you give us a preview of the talk you’ll be giving at the Amplify Festival in Australia?

I’m going to be talking about what makes self-control hard, and share some solutions that people have come up with to level the playing field between the present and the future self and show some of the technologies that my colleagues and I have created to assist in these decisions. Some of these I’ve talked about before, as well as some new ones that we’ve been developing since then.