Welcome to this episode. I'm your host, Angeline Corvaglia, and I'm here with Bill Schmarzo, a.k.a. the Dean of Big Data, and Data Girl and Friends very own Data Dean. I really love talking to you, Bill, because you're not only an expert on AI and big data, but you're also an endless optimist.

[00:00:29] And every time we go down rabbit holes of how bad things are and how risky things are, you've always managed to turn around the conversation. And since I've been really focused on the harms of, potential and real harms of AI chatbots, and I've also seen awareness is increasing. I've also seen a lot of parents saying this is another thing we have to worry about.

I'd like to talk about in general AI. How can we As you say often, use AI [00:01:00] to help us against the harms of AI.

Well, there are a lot of things, by the way Angeline, thanks for having me. I know I love these conversations, they're always fun. You ask the correct questions and together we kind of hopefully move this, like I said, in a very positive direction.

Let's focus in on what parents can do and what we can do. We were talking ahead of time about how I'm teaching a class to a bunch of high schoolers tomorrow, about how do you use a tool like ChatGPT in a more responsible way. And there's a couple of things you have to do, right? First off, we're going to go through an exercise to determine the viability of small nuclear reactors

[00:01:39] as an option for climate change challenges and clean energy. Right? This is a great tool to have that kind of conversation, but we have to prime the pump first. So the students are going to gather a bunch of different research from different academic institutes and other places, research facilities, and they're going to load it into their [00:02:00] ChatGPT first.

They're going to load it into their chat, right? You can upload it. And then with that as a frame, they can now start asking questions. Referring back to, you know, using information I gave you, what are the benefits? What are the risks? What are the pros and cons? And then teach the students that as you're having this conversation, you are also training the tool.

[00:02:23] So once you get the pros and cons, ask it, well, what's your rationale for those pros and cons? Give me your sources. And where do you think you might be wrong? Right? What else did you not consider? How accurate do you think you are? Right? So you, you have to constantly challenge it. Ask it for its rationale. Challenge your rationale. Ask it to expand its rationale. Treat it like you would, I said, a five year old, but no, it's really like a, an immature research assistant who just brings you back stuff.

You ask it a question, here you go, ask it stuff, and you gotta, you gotta constantly not only ask it more questions, but you're trying to teach it. Teach that [00:03:00] research assistant what's important for the conversation you're having. Even as far as talking about, you know, what are the benefits from a society perspective? List all those and give me a rationale. What are the benefits from an economics? What are the risks?

So part of what I think we have as parents, and especially as teachers, is to not ignore and ban this tool, because if we ban it and don't teach people properly, the students are going to use it and they're going to start believing the stuff that gets fed back to them without being a doubting Thomas and asking questions and challenging it and saying, give me your rationale.

[00:03:34] What hit me is something I think about a lot is that I think as adults in general, I was just talking to someone yesterday, we should avoid, and I agreed with him, we should say adults in general, because we often look to parents and teachers. But aunts, uncles, grandparents, concerned neighbours, so I'm going to try to remember what he told me and say adults.

Adults to kind of say if we're overwhelmed with something, as you just said, take all this [00:04:00] research, put it into your own to ChatGPT and ask it. A lot of adults would be overwhelmed with that. But we need to remember that the digital generation, even if they've never done it before, they can learn it in a second, or they can even teach each other.

So it doesn't need to be a limiter, right? The adult's role can be actually to teach or help children ask the questions. Because we don't need tech to understand how to ask questions, how to probe deeper into things, right? We, it's a tool, yeah? It's a means.

[00:04:29] It's almost like we become a five-year-old, and we're interacting with it by asking, why do you say that?

Why? Keep asking why? Why? Keep making it go to that second and third level. Make it challenge itself. And that's part of what I think is a really useful framework. We're going to go old school. The Socratic method, right? Socrates taught that there are six questions that you have to ask in any situations around assumptions and perspectives.

And he taught his students to be dubious. [00:05:00] That what you're being told, you should always think it's not true. And drive down, especially decisions that are important, right, important decisions, drive down until you have enough evidence in the questioning process that you now feel comfortable, that you, that you're confident in making the best decision.

We're going to go old school, baby. We're going to become Socrates.

Perfect. That's a really perfect way of looking at it, because if you start out questioning, these chatbots… I have used Gemini, ChatGPT, and Copilot. Oh yeah, and I tried Replica this morning.

[00:05:35] That's a 30-second conversation with the Replica. The more you ask questions, the more it really demystifies. Because you realize you can actually usually convince it to do almost anything. I mean, I don't try to push legal boundaries and things, but, and that's the thing, right? If you ask questions and it doesn't all of a sudden seem like this all powerful thing. Then it shows [00:06:00] its weaknesses pretty quickly, doesn't it?

Yeah, I almost said if you force it to think, which is a wrong way to frame it. If you force it to continue to do the next level of research, right? You continue to ask it to find more information. You continue to ask it, how accurate is that? And how do you base your assessment of accuracy? And what are your sources?

[00:06:22] If you continue just to challenge and you're dubious and you challenge, as you're having that conversation, you're actually training it in the area that you're most interested in. Yeah, you can go off on tangents, but now you're really, I'm trying to really train this research assistant to help me think more thoroughly through this.

So I can ask a question such as, what did I not consider? What did I overweight? What are other stakeholders? And then I also love the idea of taking a persona. You know, what if I was Socrates? How would I answer this question? What if I was Martin Luther King? How would I answer this question? And you start, you start using different personas to take and [00:07:00] attack us from different angles.

You are forcing that tool to consider more data in the context of the problem we're trying to solve. So the data about the Kardashians and what's going on in social media, it doesn't overwhelm it because you're not concerned with that. You're not concerned with this set of problems and you're making it find more data and heap on more learnings and more insights into that particular topic.

[00:07:23] I have maybe a difficult question. I'd like to bridge a little bit into my more, most recent small obsession, is people developing emotional attachments to AI chatbots. And there's been some examples from the US. Unfortunately, a boy recently who committed suicide and, and, the last thing that he was speaking to was his chatbot. And I was just asking some legal experts about this, why this isn't banned. And I got some really nice answers. And one of them, he was talking about parasocial AI, which is a new term that scientists [00:08:00] apparently have, have now found. And it comes from research showing how these systems created one-sided emotional bonds with users.

I got this from Arnaud Engelfried, uh, just to say where I got that from. And I said, yeah, apparently the scientists are researching this now. Yeah. And obviously the, vulnerable population, which is obviously youth, is especially vulnerable to this. So how can we take what you just said and try to convince someone who is going to try out the AI chatbot in a relationship manner to consider doing that similar approach?

[00:08:40] You make an interesting comment. You talk about so many key points there. The vulnerable. The vulnerable isn't just the young or the really old. It's anybody who is not sceptical, who overall takes a look at something and wants to challenge it, is, [00:09:00] I think that our schools need to spend more time than ever on teaching our students how to think, not what to think.

Think about a lot of our problems in social media, on cable news networks. We have a lot of people who are very vulnerable to someone of authority stating something as a fact. And them accepting it. And then parroting that forward, right? Parroting is not intelligence. [00:09:28] As a society, we need AI data literacy is critical, but it has to be taught around how do we as humans make more informed decisions to improve our odds of making, of surviving and making the right decisions.

So, to me, it starts with teaching people how to think. Not what to think. I think it's part of a, a transformation, part of a challenge. You think about schools. We always told, you know, schools are all about teaching you to memorize and regurgitate. [00:10:00] Teaching you what to memorize, right, what to know. They taught you what to know.

And now we're saying, no, no, that with Gen AI, the people who are going to be successful aren't the ones who can memorize and regurgitate. They're the ones who can think and apply. And that's a total flip. It's an empowerment. And I'm not just being told what to think. I'm having a chance to experience it, to explore it, to make my own rationale. Because we know that 1 plus 1 is 2, but beyond that, about everything else we've learned about history and society is always in the grey.

[00:10:30] And how do we make rational decisions about those things that are in the grey area?

I think that's a really good point, and I love that we're making it clear that in order to do something about the current digital situation and AI, the first steps are not online at all.

It's in your head. I was just speaking to actually a university instructor last month. We talked about the future of education, right? There's all this talk about the future of [00:11:00] education. What I say is, for me, the future of education is that you have to tell the students why they're supposed to do something exactly as you say, because if they don't understand why, what they have to learn, then they'll just use a chatbot.

I mean, I would have used a chatbot. Anybody would do it. You don't have to pretend. and then she said, when I said that, she said, it's true, actually, that there's always going to be cheaters. There's always going to be people who take the shortcut, but if you tell them, If you take the shortcut, you're just not going to succeed in the long run in the world, the AI-filled [00:11:34] world.

So you, you, you, mark, we as a very positive person. And I am incredibly positive about this. I think that we are due for a fundamental reforming of what we do from an education perspective. And I think when we do that, the things that are traditional, educational systems have thwarted or hindered curiosity, imagination, exploration, [00:12:00] right? That we, those are human traits. They're built inside of each and every one of us.

But we go through classes. We take standardized classes and standardized tests and standardized classrooms and standardized textbooks. Everything that we've done in our education system is all about standardization of memorization and regurgitation. And that's what makes humans so powerful, such a powerful capability, our ability to imagine, to be curious, to explore, gets thwarted.

[00:12:26] We have a chance to transform that tremendously. AI can be one of the vehicles for doing that. Because it's got the memorization and regurgitation down. I don't need to worry about that. I'm now going to, and I like to call it my Yoda, because I've got a little Yoda sitting right back here. Hi Yoda, how are you doing?

Right? And that Yoda is Your Own Digital Assistant. I'm making it work for me. I don't have to worry about the, that I've written over 500 blogs in my life. I can ask it a bunch of questions. It can go and find all this information, my 500 blogs, bring it together. I can challenge it. I can, I can make it do all that heavy lifting work.

[00:13:00] So I get to spend more time thinking, being curious, imagining, exploring, trying different things. I'm always trying ideas out with my, My digital chat GPT out there, the Dean of Big Data GPT. I'm always throwing new ideas at it, and I'm saying, tell me what you think about this, and what's your rationale for the decision?

Is it a good idea? It's not a good idea, why? How do I make it better? So I think there's a really incredible chance for AI to make us humans more human. And to leverage that which makes us powerful entities. So I'm very encouraged by what's going to happen. Yeah, there's going to be pain points. Yeah, there's going to be people who cheat, who go around the edges.

[00:13:37] But in the long run, the people who cheat are only cheating themselves. They're not going to learn how to think. They're gonna, we're gonna be told what to think and that makes them, in my opinion, the vulnerable. The people who are very easily swayed. And we see that today through social media and cable news, but it's been a challenge in history. Propaganda you know, the Hitler and, and the Nazis, they've used it very effectively. [00:14:00] They didn't have social media, but they challenged people to not think. They just told them what to think. And Germans, by the way, Germans, really smart people are all like, Oh yeah, we think like this.

Education system failed them, and it's failing us if it doesn't refocus on making us humans more human and empowering us to challenge and be dubious and to use this tool, this Yoda, to help me nurture and fuel my curiosity and imagination.

[00:14:30] Exactly. Yeah, just also use it actually to, to help figure out what to do, right?

nd that's one thing that, that is easy to forget for people who aren't, who don't think about this and spend a lot of time on it, that it doesn't want to cheat you.

It's not that smart. It's not that smart. It's not going to teach you. It's only going to do what you teach it or train it to do. It's really not, it's not an evil force.

It's not like the bad guy where you just see. You can easily tell it what you want to do and, and [00:15:00] it will give you a kind of a neutral answer. And I think that's important to keep in mind, right? It's not the enemy. It can be used for very bad, but it can also be used for a very good. And another thing I remember the name of the book we were talking about before. It’s called Stolen Focus.

And there was one thing I read just the other day. And I think this is so perfect for everyone who's kind of panicking about how AI and AI chatbots are gonna, and social media companies are going to completely destroy society. [00:15:32] And they were talking about how when women got the right to vote. At the time, there was the power structures that were working against women having the power to vote were much stronger than today's tech companies.

And still, women were able to band together and got the right to vote. So, there's the examples in history, where multiple times… Also the another example of getting weekends [00:16:00] free in the industrial revolution. So that's another thing, right? That we can use AI to help us figure out how to do it, how to band together and make the society the way we want it.

So it doesn't have to be anti tech, right? It can be tech on our side.

The beauty of AI is that traditional analytic models have made predictions based on the trends, patterns, relationships it saw in its historical data and just projected that forward. So all the biases in your historical data are amplified in your predictions, right?

[00:16:37] You end up with confirmation bias because you're talking to the same people on the same products, the same message you end up with. Unintended consequences you've not. It's very narrow, right? Some organizations end up with a total addressable market shrink because you're shrinking the number of people talking to.

AI is different. AI allows me to define the variables and metrics around what I aspire to be. To put [00:17:00] those into the AI model, the AI utility function. So my AI models can factor in things like environmental concerns or community concerns or society concerns and fairness and ethical and responsible, right?

It can, the AI models can be constructed in a manner to do that. And historically, if your environmental data and concerns were not in your historical data, you were, you know, too bad. But now we have a chance to change that. So to me, AI is a very powerful enabler of enabling what we as humans aspire to be, not what we've been.

[00:17:35] But what we aspire to be. And so I think once we, once we start educating people, and bringing people into that process, educating people into their role, and making sure their views and perspectives are brought into this AI utility function, we will raise the fortunes of everyone. There will be a huge AI dividend. There's no reason why everybody can't win, because there's enough pie for everybody to be successful.

I think that's [00:18:00] the perfect thought to close with. And I want anyone who's listening to really spend some time, no matter how pessimistic you might be at the moment, to believe this. Because we need to believe that change is possible and positivity is possible. Because the AI is only, as you say, what we teach it to be, and we can teach it to be positive. We can take back the narrative from people who only want to make money at the expense of other people's data.

[00:18:29] We can! And we will. And we will. Exactly. We can and we will. That's right. Join the group of people trying to make, make it happen.

We're going to march. Yeah we are, absolutely. You're going to make this happen. We're going to band together worldwide. Yeah, because it's a worldwide problem, worldwide issue.

So really, I love every time I talk to you, we turn to the positive. So thank you so much for, for being here. It was great. Thanks for having me on. As I said, we always have great conversations. [00:19:00] You always make me think and explore. Lots of notes I take, as you noticed.

kay, thank you. Then talk to you soon. Alright, cheers.

Please let us know what you think about what we're talking about in this episode and the others. Check out more about us and subscribe at digi-dominoes. com. Thank you so much for listening. I'd also like to thank our sponsor, Data Girl and Friends. Their mission is to build awareness and foster critical thinking about AI, online safety, privacy, and digital citizenship through fun, engaging, and informative content.

[00:19:36] Check them out at data-girl-and-friends.com. Until next time, stay curious and keep learning. Digital Dominoes!