Artificial intelligence (AI) is a topic that sparks many debates. Everyone is curious about the future of AI, whether it can generate original ideas, and if it will ever replace humans. But the crucial question we should be asking is, how we can best work with AI to augment our own capabilities.
We dive into this and so much more with a data analytics executive, Glenn Wasson. Glenn is a data scientist and the Administrator of Analytics and Performance Measurement at the University of Virginia Health System. He also holds a PhD in Computer Science from the University of Virginia. Glenn has leveraged his expertise to creatively solve problems across diverse industries, including healthcare, national defense, energy, and higher education.
In this episode of the Scaling Tech podcast, Arin and Glenn explore the capabilities of AI in depth. They highlight key features of AI, examine its main limitations, and share practical advice for making the most out of AI in an ever-evolving tech space.
Check out this episode to stay ahead of the curve and master the essentials of the new business norm that is AI.
Listen on Spotify
Listen on Apple Podcasts
Watch the video:
Key Insights with links to jump ahead are below
About Guest
Name: Glenn Wasson
What he does: He’s a Data Analytics Executive
Company: UVA Health
Where to find Glenn: LinkedIn
Key Insights
⚡AI reminds us of us. Artificial intelligence gets its name because it mirrors human behavior. It reminds us of us, and that’s one of its greatest strengths — it behaves in a natural and easy-to-understand way, ultimately creating a better user experience.
Glenn explains, “AI is this big umbrella term for systems that solve problems in a way that reminds us of us. So, in a way that feels human-like to other people. It’s a bit of a philosophical definition, but it sort of avoids having to define exactly what intelligence is, which is a notoriously sticky issue. We don’t have to argue about whether AI or people… It allows us to ascribe intelligence to what systems we want, and that’s not so much sort of diss of the state of the art because the important thing there is how easy it is for someone to interact with a system that’s sort of acts the way they would anticípate a person doing. So you can really get a better user experience, let’s say, out of such systems.”
⚡Machine learning (ML) is a context-dependent probability model. While AI mimics human behavior, ML is more powerful because it learns from examples rather than relying on predefined rules.
Glenn explains, “We talked about AI as this big umbrella term. Machine learning, then is a subset of AI where we’re creating systems that try to determine patterns from examples. So really useful in situations where it’s hard to write the algorithm or the ruleset that governors how to solve a problem, but we can differently judge any particular solution to be right, wrong, good, bad, and so if you can present a whole bunch of positive negative examples to a system that can try and derive what the rules are from that, that’s machine learning in a nutshell. And then a large language model, a particular technique where by we train essentially this giant probability machine on all the text we can find, and its job is to say, given all the words that have been used up to now, what’s the most likely next word?”
⚡Can AI build on our ideas? The boundaries of AI creativity are a topic of ongoing debate. Glenn shares his point of view. “There’s a theory of creativity where what is really happening is people are combining parts of things they have seen before just in an unusual way. So they are reusing things that are already out there, which the AI, in theory, could have access to as well, and so it’s just a question of whether the LLM technique will cause it to combine those pieces in an unexpected way or will it do the most expected thing because the probability model said that that’s what it should do?”
Episode Highlights
What are the limitations of AI creativity?
While AI can genuinely improve human capabilities, the ongoing debate is whether it can generate truly original ideas.
Glenn says, “This is a great philosophical question, too. Can it create? We tend to think of the answers that we get back from ChatGPT or Claude or any of these things that’s sort of some kind of average of all the things to have been said on a particular topic. The system doesn’t really understand in any deep way what you’re asking, but it knows what the most likely response is which means it’s going to be something that somebody said before or some amalgam. That can be really useful in a domain where you don’t have a lot of expertise, and you need the summary. But it really depends on how you want to define creativity whether that system could ever create something new. And certainly, when it’s it’s trying to create the average, if you will, of all of these things, it’s probably not pushing the boundaries, which tends to be where creativity is.”
How can AI and people collaborate most effectively?
AI can improve human decision-making by providing valuable support without taking over the process.
Arin shares his thoughts, “It’s not only helping the doctor potentially get their hands off the keyboard and be more attentive to the patient but then, without giving a diagnosis, you still open the possibility there that it will assist the doctor by saying, ‘Well, they talked about these things so this is the most common prescription for that’ or ‘These are diagnoses you should consider.’ It’s not telling what the diagnosis is. It’s not writing a script for the prescription but it may help prompt the doctor.”
Glenn adds, “That is kind of another level of decision support, the sort of lowest level, it’s just the summary of what was talked about. They said that they have this problem. They experienced it in this way. They have this level of pain. They have these sorts of social determinants. And the physician can still turn that into a diagnosis, but it was helpful just to boil the ocean down to those facts.”
How to encourage innovation in a constantly evolving tech field?
Setting up a development environment is crucial for driving innovation, not just in AI but across all areas of technology.
Glenn says, “Having a dev environment is maybe one of the most important things you can do, and then for people who are kind of trying to bootstrap themselves, we’ve been working in AI for a long time, but the large language models is sort of a new thing, and we didn’t have people who are kind of dyed in the wool on that, and so you have to figure out how to bootstrap enough knowledge that you can get to that experimental phase where you can sort of figure out what can I make this do for me, and that can be really hard if you’re just reading a bunch of blog posts or whatever. It’s an old school way but I found Coursera to be pretty useful on this in terms of kind of summarizing the state of that, but also reminding you of things that we knew about AI long back that still apply.”