Alexis
Hi everyone, welcome to Talk Tech with Data Dave. I’m Alexis, your host of this podcast, and I’m here with my dear friend, Data Dave, to talk about all things data, all things cloud, all things technology, and all things D3Clarity. Hey, Dave. 

Data Dave
Hey Alexis, how are you? 

Alexis
I’m good. I’m getting excited about this podcast today because we got a listener question.  

Data Dave
Oh, excellent.  

Alexis
I love listener questions, and if you have a question for Data Dave, feel free to e-mail us at talktech@d3clarity.com, and we’d love to answer your question on the podcast. 

All right, Dave. So, today, my question comes from one of my friends on LinkedIn. She works at an institution of higher education. So, artificial intelligence has a really big impact on higher education, like specifically ChatGPT. The message she sent me said, “Hey, Alexis, I know a lot of people are nervous about artificial intelligence. Can you guys talk about AI and how we’re already using it in our everyday life?”  

So, Dave, I was hoping today we could kind of break that into two questions. We could talk about artificial intelligence a little bit, talk about AI a little bit and then talk about how we are already using AI in our everyday lives. 

Data Dave
Absolutely. I think that’s a good way to break it down. If we talk a little bit about what it is, then just how intelligent is it? 

Alexis
Yeah, for sure. So that’s the first question. What is AI? What is artificial intelligence? 

Data Dave
So, artificial intelligence is fundamental, and if you look it up, you probably get all kinds of definitions. But the artificial intelligence is really… take the 2 words. Artificial is man-made or not natural, and intelligence is reasoning. Essentially, it’s the reasoning and inference of information from the information that’s around us, the data that’s around us.  

So, artificial intelligence is the ability to infer an outcome or mimic human intelligence if you like, with something that is not human. That is not natural. That is artificial.  

Whenever we make a prediction with a computer, we’re essentially doing artificial intelligence. We’re taking a set of data and inferring an outcome from that data. That can be as intelligent or as not intelligent or as dumb as it might be. That’s the fundamentals. 

Now the modern wave of artificial intelligence is the sort of true mimicking of intelligence and the ability to learn and the ability to infer knowledge and depth of knowledge using more advanced mathematical decision trees and things like that, within a cadre of data. So, the prediction gets less obvious, if you like, and more as they say pseudo intelligent in terms of how it was reached. 

Alexis
You just said the ability to learn. That is what I think about when I think about Terminator.  

Data Dave
Yep, Yep.  

Alexis
That idea of artificial intelligence – where all of a sudden the technology has the ability to learn, and that’s what I think it scares a lot of people about AI. But whatever we talk about it from a destroying human point of view, that makes sense. But when we talk about it from an analytics point of view, like, just like a data point of view, it’s actually really, really helpful to be able to have the fake stuff learn as we go, right? 

Data Dave
And that’s exactly right. So, the ability to learn is a key piece. The ability to learn from your actions or from your previous experience. But you have to be told whether it’s right or wrong.  

So, you might make a prediction, and a computer might make a prediction. And if that prediction is confirmed by the individual, then that adds weighting or adds construct to that decision being correct. And so, it’s going to lean the algorithm then with computer learning, towards making that decision again in the future.  

And that’s similar to what you would do, right? If you were to say, “Hey, Dave, I think you’re going to wear a blue shirt on the next podcast.” Then, it proves to be correct. You’re more likely to predict that I’m gonna wear a blue shirt again, as opposed to a red one. Or a green one, right? 

So, when we’re talking about learning, it’s putting in the algorithm, that’s traditional computer programs and computer structures would say. “Based off of just mathematics, in the rainbow of colors, there’s one in seven chance that Dave is going to wear a blue shirt.” Well, no, there’s a whole bunch of variables that are going to predict that I’m likely to wear a blue shirt. First of all, it’s the only shirts I’ve got with D3Clarity written on them. I’m not blue, so I’m likely to wear a blue shirt. Blue, as you can probably see from behind us, and your shirt is our colors. It’s more likely that I’m gonna wear a blue shirt for this podcast. So when we bring in that aspect that becomes more likely. But still, fairly deterministic with the set of variables. When we say, “Well, last time he wore a blue shirt, and he’s worn blue shirts in the past, probably 80% of the time.” Then we’re starting to use more machine learning to start to say, ”Well, we made the prediction, did the prediction come true? Did we learn from the fact that it did? Or did not come true.” 

Alexis
That is great. I definitely followed you, but I want to follow back on something you did just say. You said machine learning, so I hear AI and machine learning, artificial intelligence and machine learning often kind of being used interchangeably, and I feel like maybe you just did as well. Could you explain if there what is the difference, or if I can use them kind of interchangeably? 

Data Dave
You can and you can’t. There are a lot of gray areas in a lot of these things, AI is artificial intelligence, is the ability to mimic human intelligence or intelligent decision making using a computer or using machine? AI does not necessarily have to learn. So, you can have AI without having machine learning. You can have machine learning without having AI. 

Alexis
But they are often used together. 

Data Dave
But they do often go hand to hand. If you have AI without machine learning, you’ve got a complex algorithm or a neural network, or a complex decision tree that will always pass the same way for the same outcome. But when you then say, “Yeah, but I’m actually gonna engineer the fact that if I get it wrong, I’m gonna change my outcome or I’m gonna update my data set.” And so the computer is learning based off of its interaction. 

Alexis
That’s where machine learning comes in. 

Data Dave
And you have to teach the machine to do certain things and that sort of thing. The classic example of sort of machine learning and AI together is natural language processing. When you talk to a machine. You actually confirm, and it learns your voice patterns, your vocabulary, and its vocabulary increases as you talk to it. As you do when you hear a new word. You look up the meaning, and you then use it in a sentence. 

Alexis
Or I ask Data Dave about it. 

Data Dave
Exactly! With machine learning, it is very similar when we use AI for natural language processing that is often machine learning as well. You have to teach the machine the vocabulary. You have to teach the machine the grammar, and then it will reflect it back to you. And a common place where we use that all the time is in chatbots. When you type in and get customer support from your insurance company or your bank, or your whatever that is, a lot of AI and natural language support, etcetera. 

Alexis
I think I’m teaching Microsoft how to understand us better because I’ve been transcribing all of our podcasts in Microsoft Word. And it’s slowly understanding our voices better and better and better. And our last transcript was beautiful. It was so easy to do because it finally understood us because it’s gone through our recording so many times. That’s machine learning.  

Data Dave
Yes, that’s exactly machine learning. You’re teaching it. You’re teaching it the vocabulary that you use and the voice patterns and the sentence structure that we use, and that gets quite difficult. It’s quite complicated. Especially for somebody like me who’s got a weird accent and an even weirder sort of central Atlantic pattern of sentence construction. 

Alexis
This was my favorite. It transcribed you saying behavior into the English spelling of the word behavior. 

Data Dave
Interesting. That’s very interest. 

Alexis
Isn’t that crazy? 

Data Dave
That is crazy, yes. 

Alexis
I saw that on my last transcript, and I was like, “It understands Dave’s accent. That’s amazing.” Yeah, I’m sorry. I got a side track. You were talking about chatbots. I was just using a chatbot today, talking to an insurance and it was a little bit confusing. We looped a lot of times before I found the right answer. 

Data Dave
They are using AI to structure that chat to be as close to a human chat as possible because otherwise, it becomes artificial, and you will abandon it right away. There’s a lot of that sort of stuff and it’s really that complex inference and AI and machine learning becomes… We’re evolving better algorithms for learning regardless of the subject. The idea of using a complex decision tree or a neural network or machine learning to present something and  be told it’s wrong and know that it’s wrong and present something else next time regardless of the subject. That can then lead to a learning construct and an artificially intelligent outcome, which is why we’re seeing this massive explosion at the moment. It’s the idea that there are common algorithms that can be used for machine learning and AI, which I can now apply to multiple different subjects. So, whether it’s car maintenance or the weather or different things, I can actually use similar constructs to get there. 

Alexis
And that leads into kind of the second-half of the question, which was, “What are some ways that we are using AI already in our everyday lives?”  

So, you just used the example of a chatbot, which I’m sure we’ve all used at least once before, but are there other examples of the use of AI that we’re already doing and maybe even don’t realize it. 

Data Dave
Almost certainly, most of the prediction algorithms for certainly some of the big entertainment platforms are all AI based. 

Alexis 

Ohh so like if I get on Netflix, not a sponsor. Although Netflix, if you want to sponsor Talk Tech with Data Dave, we would be happy to have you.  

But when I get on Netflix, it’s like, “Hey, you probably like romantic comedies,” because I’m a classic girl who likes romantic comedies, and I watch a lot of them. 

Data Dave
Exactly. So there’s a lot of that sort of stuff, and the algorithms are getting increasingly more complex using AI to make these predictions. So, that’s another one where any of those sorts of predictions, the ads that Google or Amazon place in front of you are almost certainly getting more and more robust in that regard and starting to learn with more and more variables. And really, that’s the idea. Because you’ve built something that can now learn and can now work with a lot of variables, the number of variables can grow exponentially.  

If you have to code the algorithm for prediction yourself, that becomes very finite. If you can allow it to learn. You can feed in variables that you didn’t necessarily know about before because it can compute through a lot more variables. The weather might impact whether I’m wearing a blue shirt or not. If you consider the weather in Texas when we’re doing a podcast, you can probably predict whether I’m going to wear a blue. And whether it’s blue or not, you might not be able to do, but the fact that it’s a short sleeve shirt you can predict. 

Alexis
Yes, because you’re a good podcaster and you prepare and you wear D3Clarity clothes for the podcast. I’m not so much, although today I got on our new one and I’m very excited about it. I’m usually in like sweaters because it’s cold here where I live, but yeah, okay.  

So, the idea of a machine or a computer being able to figure out things like that and learning based on other variables, that’s the essence of machine learning.  

Data Dave
Right exactly. Now, another areas that use it all the time is probably every time you ask for directions from your smartphone. 

Alexis
Oh, which is often. 

Data Dave
Which is often. Because if you think about the amount of alternatives for a route between two places can be pretty huge, and that’s a pretty complex algorithm. When they then include traffic situations, accidents, police presence, etcetera to give you the most optimal route, that is probably an AI-based algorithm. Almost certainly. 

Alexis
Because it’s definitely not somebody in the background looking at the maps going, “You should take this route, Alexis.” That would be impossible. 

Data Dave
Right, it’s not exactly. And not somebody sending you in circles and so on. It’s not.  

And it’s learning from the maps being updated and various other things without humans being involved in the intelligence to get you from A to B.  

Game theory, playing games and mimicking people. For gamers all the time. That’s a lot of AI and classic AI. In fact, some of the AI algorithms that we use for business came from the gaming community as we evolved that side of it. And then there’s things like facial recognition and other software in niche areas like that. 

Alexis
So, the big picture is… we’re likely already using AI and machine learning in a lot of different places in our life that we might not even realize we’re using it. And it may or may not be something to be worried about. That’s a different conversation, a different, maybe more philosophical conversation. 

Data Dave
Well, yeah, I think we can go into more depth in another chat in another session chat session, right?  

I’ll be a bot next time. Then we could just have a chat session. 

But the thought I would leave people with is not so much the algorithm. It’s do we trust the machine and why is the machine more or less trustable than the person? The machine, to a certain extent, is more predictable. But is it predictable if we don’t know what data it’s looking at to make these determinations and make these predictions? 

From a philosophical point of view, I guess it gets to the definition of trust as much as anything else, which is, “Why do you trust this piece of information? And do you trust that the information you’re receiving isn’t biased in some way, and do you expect it to be biased?” is another question.  

If somebody is trying to sell you something, whether it’s a human making that prediction or a machine making that prediction, the odds are they’re going to predict that you should buy it, right? And so that’s an element of bias, and we have to go into these interactions expecting that level of bias, and we as a society need to sort of learn how to interact and how to use our AI and how to police it. The same as we don’t naturally trust everything everybody says, we can’t trust everything that a computer says. 

Alexis
And that leads me back to a conversation we had during the “Where is Data Going?” podcast where you pointed out that if you feed data into AI and AI gives you a bad answer, that’s not AI’s fault. That’s my fault. I fed it bad data, so I have to take responsibility for my data, and that was a really great point that you made during that session that I think perfectly lines up with this conversation we’re talking about. 

Data Dave
That’s right. And now, you can get bias from an AI or the AI algorithm and the data work in conjunction to give you a bad output. So, there is the concept of bad AI if you like, and good AI. That sounds so flippant to a certain extent. But the idea that you have malicious AI versus good AI. It’s the same with people. Every prediction has an agenda. If it’s being honest, whether I predict that you’re wearing a blue shirt, I’ve got an agenda for why I’m predicting that you’re gonna wear a blue shirt. 

Alexis
Right. She wants me to wear a blue shirt. 

Data Dave
Are we going to match? I’m going to match. Do I want you to wear a blue shirt? Why not want you to wear a blue shirt? How am I going to incent you to wear a blue shirt in the future based off of whether you did or not. 

Alexis
And the machine will do the same thing. 

Data Dave
The machine will do the same thing based off of what the organization that owns it wants you to do.  

So, you have to know that, and you have to weigh its answer within that construct. If you use a chatbot from an insurance company, the odds are, it’s going to recommend that you switch to that insurance company. If you ask for rates from that insurance company, it’s probably going to come up with good rates for you, but that’s the same when you’re talking to a real person as well. 

Alexis
Right. That was a good wrap-up there, Dave. That was a good conversation. I think that I have a better understanding of the things that I use AI for. I don’t even think I had put the idea of like my Google Maps as AI prior to this conversation, so that’s really helpful. I do think we should have that philosophical conversation of good AI versus bad AI, but probably not today. 

Data Dave
Probably not today, yeah. 

Alexis
But thank you everyone for joining us. And if you have a question for Data Dave, please feel free to e-mail us at talktech@d3clarity.com or you can submit a question right on our web page. Dave, thanks for talking with me today. As always, it’s been a pleasure. 

Data Dave
Thank you. And thank you everybody for listening and thank you Alexis, for hosting us as usual. 

Alexis
Thanks, everyone. Bye.