Murray Scott: There is a general suspicion sometimes about artificial intelligence. For you, what are the main benefits of AI and why should people be excited and positive about the impact of AI on society?
Bill Schmarzo: The positive side is our ability to do more with less. Using AI we can make very precise decisions across a wide range of cases – everything from business, to healthcare, to education, to society, to diversity issues, to environmental issues. We can start to move from making overly generalized policy and operational decisions that have led to lots of inequities in society and move to an environment where we’re making very precise decisions that help people, but actually cost less.
To me, that is the exciting thing. We can start transitioning from decisions based on averages to decisions based on predicted propensities. That’s a hard shift in how people think about things but the ramifications are staggering.
Murray: An important challenge is educating people in how we use data and bring data into those kinds of scenarios when making big decisions. What are the some of those challenges in developing a responsible type of AI?
Bill: There is a lot of discussion regarding data literacy and data literacy has two components. There is a data awareness component and then an AI literacy component. From a data-awareness perspective, people need to understand how their data is being used or how it could be used. They need to be very vigilant in how they protect their own data. We don’t do that very well today, and so there needs to be much more of an awareness effort put in place to ensure that personal data is not being used for nefarious reasons.
We kind of gloss over this issue. We don’t even talk about our data footprint and people don’t understand, when they use their iPhones, how much data is being captured by these devices. So, there needs to be education and awareness on data, but then there needs to be this conversation and education on AI.
What makes AI both powerful and scary, is this ability to continuously learn and adapt with minimal if any human intervention. The challenge becomes, in that kind of environment where the AI can learn faster than any human could ever learn, is that you have to be very careful about how you define the variables around which you want to drive that learning.
Murray: AI is incredibly powerful but surely humans drive and control that power?
Bill: I would argue that AI is not smart at all. AI is really stupid. Given the variables you tell it to optimise against, it’s just going to continue to learn and try different things. And you’re spot on. We think that AI is the is the realm of the data scientists and I argue that AI is a realm of being human. What AI is ultimately going to do, is force humans to become more human.