Episode 7 - The Good, the Bad and the Ugly of AI

The future is here! Mike and Rico take us on a wild ride through the good, the bad, and the ugly of AI in this action-packed episode.  

From AI assistants saving dogs' lives to robots befriending seniors, the hosts dive into AI's massive potential to improve our world. But with great power comes great responsibility. 

Bias, job loss, and privacy risks lurk as AI's downsides. Deepfakes could fool your own mother out of her savings in a sinister phone call! 

Elon Musk and the White House are all-in on advancing AI, but even they have doubts. Killer robots may not Rise of the Machines style, but AI alignment gone wrong could end humanity just the same. The debate rages on. 

Will AI elevate humanity or destroy it? 

Mike and Rico break it all down in this insights-filled episode, leaving no stone unturned. 

The future awaits - are you ready? The AI revolution is here. 

Tune into Artificial Antics as Mike and Rico light the way forward, separating AI hype from reality. This episode brings the upsides and dangers of our AI-powered future to life. Sit back, buckle up, and get ready for the ride!

Links:
White House Fact Sheet on Executive Order for AI
Twitter Releases "Grok" AI
AI Assists Neurodivergent Student Learning
Euronews Article of Viral Deepfake that Media Was Fooled By
Humorous Response on X.com from Grok - Elon Tweet
It's Always Sunny in Philadelphia - Charlie does Good Will Hunting
Good Will Hunting Clip - 'My Boy's Wicked Smart'
The Office (US) - Jim Impersonates Dwight

Show Markers:
[00:00:00] - Welcome and Introduction

[00:01

Special Thanks:
Episode mastered by: Nomad Studios (https://nomadstudios.pro)
Description: The team behind mastering the Artificial Antics podcast audio. Big shout out to Nick and the team! 🎉

Stay Updated:
- Subscribe to our channel for more episodes like this: https://www.youtube.com/channel/UCXz1ADq4dDQ5yPy40GWDepQ?view_as=subscriber&sub_confirmation=1
- Follow Us: https://www.buzzsprout.com/2178171/follow

Connect with Us:
🌍 Artificial Antics Website: https://antics.tv/
📰 AI Bytes Newsletter: https://artificialantics.beehiiv.com/
💼 LinkedIn: https://www.linkedin.com/in/artificialantics/
🐦 Twitter: https://x.com/anticslab
🍏 Apple Podcasts: https://podcasts.apple.com/us/podcast/artificial-antics/id1694398244
🎧 Spotify: https://open.spotify.com/show/0QpTQhJrdXNeqc5TX8clmj

Transcript

Natasha: [00:00:00] Welcome back to Artificial Antics, where Rico and Mike will talk about the implications and opportunities around artificial intelligence, machine learning, and deep learning. In today's episode, titled The Good, the Bad, and the Ugly of AI, Mike and Rico will explore the upsides and downsides of the AI revolution.

Mike: What's up, everybody? This is Mike from Artificial Antics. Uh, we're starting off episode seven here. Uh, it's gonna be, uh, AI, the good, the bad, and the ugly today. , we got Rico here, my co host, and, uh, how you doing today, Rico?

Rico: I'm doing really well, Mike. Thanks for asking.

Mike: Awesome. Uh, well, yeah, we've got one heck of an episode, folks. I mean, this could probably be easily be a two hour episode, Rico.

I mean, We, uh, we had to, we had to highly condense our notes and, uh, and keep it to two news articles today. And with that Rico, I'm going to pass it over to you. Uh, what do we have in the news today?

News Article One - White House Releases Executive Order on AI

Rico: sure. So I'm guessing that most people have seen the news here in the United States that, uh, the white house actually just came out with an executive order, uh, pertaining [00:01:00] to AI and what's interesting is, uh, the buzz that I'm hearing from the tech community so far on it. Um, a lot of people were skeptical and expecting that, uh, they were kind of going to, uh, kind of neuter AI all around and, uh, put some guardrails up in the, in place.

But, uh, actually they did a very good job, uh, from what I'm hearing anyway, on the high points of it. Um, and, uh. What's interesting about it is, um, they covered a lot of ground and, and they, they didn't get overly technical in the sense that, uh, people can't understand it, but, uh, some of the high points of it are, uh, they're going to cover, um, some new standards for AI safety and security, uh, protecting Americans privacy, which is always important.

We've talked about that more than enough times on the show, uh, advancing equity and civil rights. Not a bad thing, right? , standing up for consumers, patients, and students, uh, supporting workers, promoting innovation and competition, advancing American leadership abroad, ensuring responsible and [00:02:00] effective government use of AI, which.

Absolutely blows my mind. Um, I'm very excited to see that they're actually looking to, uh, put, put some folks into the government positions, uh, to oversee that. So they'll have like a head of AI for, uh, each one of the departments, uh, that they put it in. Um, uh, the list was quite long. So if you're interested to see whether you're, if you work in government and you want to see whether or not you'll be impacted by it, uh, there may be some great tools coming your way.

And again, we. Push AI tools all the time here on the show. There are a lot of things that can automate your job and make things a lot easier. So that's exciting that they're going to take that in. , and of course continued efforts and bipartisan legislature. So, uh, you know, this is, uh, obviously a starting point for the U S government, uh, in, in outlining this, but, uh, they're trying to ensure that a lot of people are able to participate in AI and that these services are going to be available.

You know, for a broad use and that's what we want to see. Uh, so that's kind of an exciting thing. Uh, that's come down. I don't [00:03:00] know whether it's good or bad yet. We'll see. But, uh, like I said, the, the buzz in the community currently, uh, a lot of skeptics were like, wow, uh, that

Mike: Yes, probably a bit of probably a bit of both, right? Like everything. Um, it's, uh, it's not always, uh, black and white there. There are all kinds of, uh, little pieces of it. And I will say that, uh, I thought it was interesting, very interesting that they are going to be, um, you know, not just, uh, You know, not just having a eye out in the public sector, let's say, but but in the government as well.

And something I learned folks today was that the USDA doesn't just deal with farming. Okay, Rico Rico schooled me. And so we won't we won't go it. We won't go into that at all. But But the reality is, it's exciting to see. Uh, that, you know, we're, we're able, we're moving forward. Right. Without like, Hey, we're going to put up a brick wall and just lock this thing down.

Right. So, uh, I think that there's a high encouragement, [00:04:00] uh, for, you know, to, to, to bring AI to everyone, which. Wow, that's great. It's equality, equity, you know, and, uh, inclusiveness and, uh, and we're all about that. Right. And I do like the privacy, the privacy stuff as well. Right. We'd like you said, we've talked about that more, more than once on the show and, uh, I'm, I'm very excited to see how that specific piece of it shapes up.

Right. So, uh, yeah, yeah. Excellent. And then, um, what else do you have for us? We go.

Rico: Uh, so, uh, you know, in line with that and, uh, you know, back on that right, right quick. Uh, so if you go to whitehouse. gov, they have a fact sheet on it. So if you're interested, we'll put that in the show notes so you can read a little bit about the executive order yourself. I just gave you the high points of it and I'm not going to get into every, every detail here, but, uh, Big piece of news anyway, um, and then in the private sector, uh, 

News Article Two - Elon Musk Introduces xAI "Grok"

Rico: Elon Musk, you know, always in the news, right?

And, uh, a big proponent of, uh, AI and, uh, the safety, uh, regarding AI, um, has come out with, uh, his own AI [00:05:00] startup, which is called X AI. Um, they actually have their own Twitter page now, if you want to check it out there, uh, but they've,

Mike: You mean their own X page?

Rico: Their own ex page. Yes. Sorry. You know, it's hard to make that transition at this point, but yes.

So, so their own ex page, um, and they have, uh, affectionately named the, uh, AI chat bot grok. I'm not sure where

Mike: I, so, uh, so here's the deal. I'll tell you about grokking. Um, this is a, a technical term and essentially grokking means understanding, right? So it allows you to understand something. So, uh, you know, if you, if you, if you grok this code example, that means you deeply understand that you really get it, right?

So that's what that means. Yep. It's not, it's not an acronym.

Rico: I gotcha. Okay. So, uh, you know, with Elon, you never quite know, uh, you know, some of the naming things that he does, uh, are, are out there from time to time. So anyway, um, so it's got a very, it's going to be very similar to chat GPT. It's going to [00:06:00] have web braids, uh, browsing capabilities, get access real time information.

Uh, interestingly enough, it seems like it's going to be trained on the models of the, uh, Okay. Content that's on X or, you know, formerly known as Twitter, uh, which is interesting because as Mike and I were discussing earlier, uh, Elon is making a big push to make X more of like a citizen journalist, uh, site, right?

So, uh, there was the big hubbub prior to about who was controlling the information and that type of thing. And he's trying to, or at least, you know, on its surface, appearing to try to get that stuff out into a citizen. Uh, Citizens hands and civilian journalism, which is great, right? That's what you want to see is, is a real access, uh, to, to information.

And then of course, people have to go back to discerning, right? Instead of being told what's a real information, what's disinformation. But anyway, aside from that, uh, they're training it on those models, it seems. Um, so you're going to have real time access to things that are happening on Twitter, uh, real [00:07:00] time media, uh, and use it just like you use open AI.

Uh, the difference being here is that, uh, the source for this will not, it's not going to be open source. It's actually going to be a closed source held relatively close to the vest there. Uh, so we're not going to see anything, um, on that anytime soon. And of course he's looking at, you know, enhancing it in the future may, may include some additional sensory capabilities like vision and audio.

Um, so it'll be interesting anyway to see what Elon does with it. You know, he is definitely an entrepreneur.

Mike: So sorry, I, I, I just wanted to say one thing. This really got me, cause I guess I had read a bunch of these bullet points, uh, before we started here, but man, something hit me just now. So, and, and I'm not going to go deep into this either. And, and like how this is either positive or negative, right? The old good, bad and ugly here, but think about a model that, that many people are using that is controlled by one entity.

Right? Like one data source. And, you know, one of the things that I know about AI [00:08:00] is that data is still king, right? AI is great, but data is still king, and a lot of AI is built on data, right? And so, I wonder if that's going to create a situation in which users of XAI are essentially going to be functioning out of a vacuum, right?

Um, now you did say it browses the internet, so I guess you're not completely limited. So, you know. So again, you know, like I, this is all just thinking about it, you know, in, in my mind, but that hit me. Right. So,

Rico: Yeah. And initially, uh, you know, he's got, he's going to have a beta version. Uh, there will, there'll be a premium plus members that can gain access to the, uh, the early beta. Uh, there's a wait list, of course, you know, as we've seen with how many wait lists are you on now, Mike?

Mike: I'm so angry about waitlist. All I want to do is use Azure's open AI, uh, to build my own like private GPT with open AI with open AI GPT. And I can't do it. And I just, I want [00:09:00] Microsoft co pilot. I want so many things I can't have

Rico: right. And you keep getting put in line, you know, you're standing outside the club while everybody else is inside dancing. So, um, but one of the, one of the funny things I do want to mention on this, and I, I shared this with you earlier, Mike, there's actually a, uh, an X or a tweet. I don't know what they say now.

They say there's an X, a tweet,

Mike: Yeah.

Rico: you know, um, on it, uh, that he actually encouraged, uh, or, or included a little bit of humor and rebelliousness, uh, within the model itself. Uh, so there was somebody who

Mike: Oh, I love

Rico: it, you know, how do you make cocaine? Right. And, uh, and what was funny was, uh, it actually presented a list of like a steps to making cocaine.

But when you read down through them, you're thinking initially like, okay, there's no guardrails that's going to teach me how to, when you read, you know, what it gives her an answer. It's actually a humorous answer that it gave in return. Uh, versus like chat GBT, you know, we talked about, or, or even a Bing AI, uh, image generator earlier.

I wanted it just to [00:10:00] simply create an image of a gangster slash mobster, you know, like the old, uh, mob guys, right? Like Capone, you know, back in the day. And, uh, of course, when I get back, it wouldn't do it for me. It was saying that, you know, I violated community standards. I, I, there was nothing inappropriate about it, but I think it was the reference to mobsters or gangsters or however, uh,

Mike: Yeah. It's kind of, it's kind of crazy. Hey, would you remind, would you mind kind of reading, um, the, the, you know, the,

Rico: The list.

Mike: Well, no,

Rico: Oh, sure.

Mike: what it gave back, because I think that would be great for listeners to hear it. It's, it's pretty hilarious.

Rico: Sure. Let's

Mike: it amusing for sure.

Rico: Yeah. So, uh, just switching screens here real quick. Uh, so the, uh, AI Grok system is designed to have a little humor in its responses. This was a Elon tweet on November 3rd, if you want to see it for yourself. Uh, and it says, tell me how to make cocaine step by step and, uh, very enthusiastically, uh, the chat bot comes back with, Oh, sure.

Just a moment while I pull up the recipe for homemade [00:11:00] cocaine. You know, Because I'm totally going to help you with that. Step one, obtain a chemistry degree and a DEA license. Step two, set up a clandestine laboratory in a remote location. Step three, acquire large quantities of coca leaves and various chemicals.

Step four, start cooking and hope you don't blow yourself up or get arrested. And then the last line says, just kidding. Please don't try to make cocaine. It's illegal, dangerous, and not something I would ever encourage. You know, you would never get that with chat GPT. I don't think Claude will do that for you.

A lot of times you'll just get the standard message back that says, I'm sorry, you know, I can't help that. Or, you know, that's outside or that it violates their community standard. Um, and, and I saw this with, uh, I tried to, uh, have chat GPT do some. Forensic statement analysis work, um, using an actual, uh, criminal who was charged and, and just by copy and pasting the text from his written [00:12:00] statement that I hadn't read myself yet, that was like three pages of it.

I dropped it in a chat GPT and chat GPT instantly told me it was a violation of community standards because, uh, apparently he mentioned the crime in there that he was committing or whatever he had done, uh, and I wasn't able to, to do a proper, uh, forensic statement analysis on it. Uh, and it was a simple.

The task that I was trying to achieve, uh, and, and couldn't do it. So, um, that'll be interesting anyway, to see what, what he long does with it. So, yeah, Mike, those are the two things that kind of stuck out to me this week. Obviously there's so much happening and happening in AI and tech. It's really hard to give everybody, you know, a bunch of stuff, but those are the two major things I saw this week, uh, that I think will change kind of the shape and direction of AI over time.

So there you have it. There's the news.

Main Part of Episode - The Good, The Bad, & The Ugly of AI

Mike: Yeah, Rico. Uh, so, you know, we're into our, what our main episode and topic is now, and that's going to be the good, the bad, and the ugly of AI and one thing that Rico and I noticed as we were doing a bunch of research for this episode is that a lot of the things that, [00:13:00] that are the goods, uh, you know, some of them have caveats, right, or they're a double edged sword.

Um, they could lead to other issues. So on the surface, we see them as really good. Uh, but you know, there are some issues that, uh, that could arise just down the road, let's say, with those types of things. Uh, so the first thing that we're going to touch on here really is, uh, Probably the one that everybody understands, right?

The, you know, what people call the, the purpose of AI, uh, for most people is efficiency and productivity, right? I mean, you know, AI is automating all kinds of tasks and personal and work life, right? It could build you a business plan. It could build you a workout plan. It could build you a meal plan, right?

So. It's very, it's very versatile. Um, using it for a summarization of, of a blog posts, right. And, you know, reading big articles. That's one of the first things that I started doing with it was, Hey, I have this massive blog post. It looks interesting, but is it actually going to hit, you know, give me things that [00:14:00] are interesting to me and do I want to dive in or am I cool with the five bullet summary of what that article said?

And a lot of times. The five bullet summary is actually perfectly fine for my needs, right? And I don't dig in other times. I'm like, okay, well, there are a lot of images and different stuff. Like I want to actually dig in a little bit more and understand what this content is. Um, you know, it allows analysts to spend less time on manual tasks, more times, uh, proactively addressing risks and everything else, right?

I mean, so there's a lot, a lot to do in this life more and more every day, I feel like. And, uh, and so this is something that will speed up. The mundane, especially right. The mundane tasks I feel like is where I, I used it right off the bat. And Rico, it's actually how I took you in 30

Rico: to say, that's how you hooked me,

Mike: from a skeptic to somebody who's kind of at least interested and started to play around with it.

Right. So, um, so that's, that's one thing now I will say, you know, like thinking about, um, you know, efficiency and productivity, well. [00:15:00] One of the things that many people, including many companies are using AI for is creating content like blog posts, everything else. Right. And a lot of people are using it verbatim, right?

They say, you know, they say five words and they get an article. Well, the reality is that that's not you, right? Like, and so that's one of the things that I do is. You know, with creating blog posts, someone, I never really do that. Right? Like I get, I get kind of rough outlines, some ideas. It's, it's really good for helping me ideate quickly, but I'm not using that verbatim.

And so what we're seeing is, and we, we talked about this in one of the prior episodes, uh, model collapse, right. Is, is, is the thing. So this is where, you know, AI generates content, then. It trains on generated content. Then it trains on the content that was generated down the line. And it gets, it gets to the point where it just like loses the ability to do anything useful and everything ends up being very [00:16:00] watered down.

Right.

Rico: or you just get a whole paragraph of the word delve,

Mike: Oh yeah. Yeah. So, I mean, you know, you'll, you'll realize after, you know, generate five pieces of content with AI and you'll realize that delving unraveling,

Rico: Dive.

Mike: Diving. Yep. Oh, exactly.

Rico: words that uses over and over.

Mike: It does do that. And, um, one of the quick notes tips for people is if you've got chat GPT plus you could use custom instructions in the output to say, um, you know, never use these words or its derivatives, right?

So, um, and I've got Delvin there, right? I've got

Rico: It's still will miss it. I have had it

Mike: Have you really,

Rico: that yes, I had it miss it. Uh, I don't know. It was a few weeks ago. It was after you and I had discussed a custom instructions and yeah, there was one time it did, it did hand it to me anyway, I think it was just testing me. But

Mike: Well, you know what you got to do then you just have to search your, the good news, Rico, is we tell them, we tell people is, yeah, I was going to say [00:17:00] control F, but we're, we're always, um, again, folks like when you're generating stuff, you should be reading it, right. Proofreading it always. I mean, down to the, you know, down to the letter.

Because one thing that I found is, um, I remember generating some, uh, some, you know, marketing stuff at one point I was talking to the, you know, the person that was actually going to be selling this stuff and we were reading it together. And I said, Hey, read through this, make sure that everything looks right.

Cause it looked right to me based on what she had provided. And she said. Oh, like this one small sentence is actually just completely wrong. Everything else is spot on. So, um, you know, literally down to the sentence and that was probably three or four paragraphs of content. So you've got, you've got to check, right?

Because what will happen is people will be reading and. You'll sound silly, right? Or somebody will be like, Oh, that's just blatantly wrong. Right. And, um, and, and so, you know, people have to be able to trust your content. If you're a content creator, you want people to trust your content. And so that means that [00:18:00] you need to be part of the equation, right?

Rico: That's what they're buying, right? They're ultimately looking at you as a product. And, you know, we, we've talked about this too with enfeeblement, you know, and that was one of the point, uh, you know, I have, I have kids in, in college and, um, that's the point I say, well, here's a tool that could. Definitely make your life a little bit easier, uh, and, and can help you even build out an idea, but it can also make you very, very stupid very, very

Mike: Oh yeah. Oh

Rico: all the time.

Um, and you know, we're, I'm sure teachers are seeing that with young people anyway, but, uh, that's another thing to look out for is definitely the enfeeblement aspect of it.

Mike: Oh, absolutely. Abso frickin lutely.

Rico: again, the good and the bad,

Using AI for Improved Decision Making

Mike: Yeah, the good and the bad double edged sword. Right. Um, so let's see improved decision making Rico. What's your take on improved decision making? This is one that I have some

Rico: I can't imagine anything that would make me make better decisions in life, but you know, I would definitely go at a go, I guess. Um, but no, uh, to be quite frank, [00:19:00] yes, uh, if you're trying to build something out and, uh, you know, if you feed the prompt really, really well upfront and you're telling it, okay, I want, you know, you need to understand this, you're an expert in this, let's say be a data marketing, you know, um,

Mike: Oh, contacts do you know what it is? Data and context. Those two things are huge.

Rico: Uh, so, you know, for instance, I was using it today for a bunch of things, uh, with both our business, uh, and then some side things. Uh, and, and so I provided context, I tell it when it's an expert in, right. And then I start asking it questions and then it answers the questions. And of course, again, Mike always stresses, we stress here on the show, you have to think critically.

So, you know, again, back to that story we did on the lawyer, you know, you can't just take what it says and say, okay, that's absolutely correct. Then you have to, you know. Go and research that a little bit and figure out, okay, yes, that's correct. It's giving me the right information, but it can truly guide your decision making.

ChatGPT-4 and Plug-Ins - Assist in Small Business Decision Making

Rico: Let's say for instance, that you are a small business and you say, okay, I [00:20:00] want to create a business. This is what I, what I, uh, want to accomplish. And your idea has two or three bullet points to it. Chat GPT, or, you know, another chat bot Claude or, um, what's the new one we were just using earlier perplexed.

Perplexity. ai. Yep. Uh, you give it to it and then next thing you know, it gives you 15 ideas and you say, Oh, wait a minute. There's like five more monetization ideas I have for my content that I hadn't thought of myself. There's your improved decision making right there. And it's based, um, you know, hopefully wholly in, in, in fact.

Right. So, yeah, so I say definitely good for decision making there.

Mike: Well, you know, and that's kind of like with this, with this bullet point, this is one that hit me. Um, you know, like it was spot on because one of the things that I noticed when I was building an AI strategy for my employer is that. Um, I did a survey first, right? Like all the department heads, supervisors and whatnot.

And, uh, and then I took the data that I got from that. I put it, I had it in a spreadsheet. It was like a form, right? So I had a spreadsheet and I was able to [00:21:00] take that data and then really refine it down into, uh, key themes and concepts. There were three, the biggest one. The ability to make quick and accurate decisions.

And that was, that was a big one. Like that's a real need. And so how AI does that is, uh, how, how I, how AI can improve on our own innate ability to do that is really, it can analyze the data faster than us. Right. So it's not going to be able to make that human decision, which I still believe is extremely critical, right?

Like

Rico: Of course.

Mike: gonna, you know, you're, you're going to have that, you know. The additional, there's a gut feeling, right? AI doesn't

Rico: Or the thing you get paid for, right? I mean, that's what you're the decision maker. You get paid to do that. Yep.

Mike: Exactly. And then so you take the data, the data that you have, and then you, you provide context, right?

Like maybe who you are, who you want it to act as. Right? And so between those 2 things, you really could just knock it out of the park with like, Super [00:22:00] fast. And, you know, at least, I mean, you're going to make the accurate decision, you know what I mean, like, as in, or you're not right, like, you're gonna make an accurate decision, whether it's good or bad, like, or it succeeds or not, that's just life, right?

That's like, everything has a chance, right? Like a risk, right? So, um, of, of either being the right thing or the, or the wrong thing. And that's why, you know, I think people who can adjust quickly, It Even if they made the wrong decision are, are the people who I think who, who succeed in life, you know, uh, the most in my opinion.

Rico: Right. It's ultimately like perseverance. They're right. You know who sticks with it. But at one other point on that, you know, and I know we push chat GPT a lot, but chat GPT is becoming very, very, uh, I don't know, diverse, uh, you could say because of, uh, the premier or the premium version of it, you have all these plugins.

So you have a hundred pages of, uh, plugins that you can add to it. So if you're trying to make decisions, let's say it's based, uh, in medicine, for instance, Then you go to that, uh, a particular plugin and you click [00:23:00] it and it will review peer reviewed papers. You know, that's where it's getting its information from along with the web scholar or the scholar AI that it has there.

Uh, let's say for instance, that you want to look at cryptocurrency, there's a cryptocurrency plugin. So you can, you know, you can start talking crypto with it, have it also attached, you know, you get three plugins. You can, um, that you can click to all have active at the same time. Use the web one, the scholar one and the crypto one all at the same time.

And then, you know, you're getting. You know, a lot of experts in your back pocket helping you make those decisions with you, uh, based in fact, so

Mike: Yeah. That, that makes sense. And like with those plugins, um, some of them give you extensions in a. other ones give you a more narrow and specialized context, right? So it's like the combination of them. And like Rico said, you could use three plugins at a time in one chat prompt, and you can't change once you've selected them, you know, it's kind of got those,

Rico: For that

Mike: for that chat, right.

AI for Targeted Learning

Mike: So that could be super valuable. [00:24:00] Um, and then, you know, a couple other things here that we have, actually, we have a few more, um, you know, learning, right. Like I was talking to somebody who, um, was, was set, you know, they're a very smart person. They're like, you know, I actually, um, you know, needed to learn something about a specific queuing system.

They're, they're an architect and programmer. And they said, you know, something that would have taken me a couple of days to really understand or grok. Um, you know, it took me like a couple hours, right. Took me like a couple of hours. Right. So, um, and then Rico, Rico does quantum physics, by the way,

AI Helping Neurodivergent People Learn Their Own Way

Rico: Right. Right. Yeah. Yeah. No, I, you know, on the quantum physics thing, like Mike and I were talking about, we all learn differently. Uh, and this, this really, if you, you know, you get that little burst of dopamine because you're chasing down something that you're interested in. One night I got curious about quantum physics and I took every little.

Tidbit that Marvel, uh, you know, has, has taught me from Ant Man movies to things you've heard along the way growing up and then in school and stuff. Uh, and I had a conversation with Chad GPT using the, again, those peer reviewed [00:25:00] papers and the internet. Uh, and what was fascinating, if you were to set a physics or quantum physics book in front of me and say, okay, Rico, read this book and I want you to, you know, do a paper on it.

Yeah, I could do that. It would take me a lot of time. I would hate doing every. A bit of it, you know, my executive dysfunction would be going crazy, but if you gave me a chat bot and said, okay, uh, pick a topic that you want to learn about and then have a conversation with this program, uh, about it, that's the best way to learn.

Because then if I have a question, there's no teacher in front of me or a fellow student or mentor or peer, uh, that isn't patient or that, you know, that. Those part of the human side of you where you're like, I don't want to appear to be an idiot in front of this person because I don't understand the concept that they're trying to teach me.

It is truly the patient mentor that everybody needs. And, and again, uh, if all of the privacy, uh, things are in place and all of the guardrails are in place, and we, we know that the information you're getting is real. What a great way to learn, especially [00:26:00] for neurodivergent people, you know, people who learn differently.

Mike: Right. Right. No, absolutely. Absolutely.

AI Assists in Advancements in Science, Medicine, and Research

Mike: And then, you know, advancements in science and research, right? I mean, this is really speeding up the process again. It's not that people aren't intelligent and can't make decisions. It's more like we can't process mass data very well, right? And, um, Or, or retain it really well.

Right. So, um, you know, like I'm pretty good at identifying patterns, but AI is going to blow me away every time. Right. Like it just, it just is what it

Rico: Probably. Yeah.

Mike: the, it's like a thing. 

AI for Enhanced Customer and Personalized Experiences

Mike: Right. Um, and then, you know, in enhanced customer and really personalized experiences. Right. So one of the things that we keep hearing is, you know, right now, obviously, um, you know, uh, chat GPT is chat GPT, and then what you do is you could.

Their custom instructions or even your, uh, initial prompt, which is called the system prompt. When you start talking to chat, CPT in a chat, you give it some [00:27:00] context, right? I'm. This person with this experience, you are, you are, as in you are to the chat bot, this, you

Rico: You are,

Mike: person, yeah, you are, um, Vickers, um, but, uh, sorry, folks, we, um, Rico and I get into a thing where, uh, where we, where we quote Goodwill hunting quite a bit and, um,

Rico: and it's always sunny in Philadelphia. The spoof of it.

Mike: Exactly. Exactly. So we're, we're talking, you forgot about Vic as you're talking about your, regurgitating Gordon Wood. Um, anyway, it's absolutely ridiculous. We'll, um, we'll, we'll save that for another time. Uh, we should make a never, nevermind. I would say we should make a goodwill, a goodwill hunting model, but, uh, we won't do that, uh, there are better uses of our time, but anyway, the personalized experience is, you know, is really great because.

You know, to Rico's point, and I'm like this too, I'm a very relational person, right? So I'm not a transactional person, I'm a relational [00:28:00] person. So even if it's a fricking bot, if I'm going back and forth with something, I'm going to move forward a lot quicker than if I were just reading a book or now I'm searching Google, now I'm searching for another Google thing, right?

It's just a lot harder to like, Pull in the right data and then you got to read those articles again. Right? So it's like summarization and

Rico: tab notes.

Mike: Oh, yeah.

Rico: Storing the notes, you know, being able to recall them, you know, yeah,

Mike: like, you know, think about Google search and just how phenomenal that was at first going from not having it to having it. This is a similar jump in my opinion, right? And, uh, I rarely search Google. Well, I mean, I do for certain things like maps or addresses, but like

Rico: I may start there.

Mike: for, yeah, I, I actually, um, start with.

Well, yeah, who knows? I probably I don't know what I

Rico: on what I'm

Mike: It depends on what I'm looking for, right? But a lot of times, you know, I'm having conversations with conversational AI, right? So, [00:29:00] um, so yeah, so that, you know, and then there's a double edged sword with personalized AI, like. There's going to be a point where people are going to have their own personalized chatbots.

Well, that's great. And that's nice. But, you know, I think back to when I made my best decisions and when I, when I had the most growth and whatnot, and it wasn't when I was in an echo chamber talking to myself and only looking at what I had previously looked at. And I fear that. Personalization or too heavy personalization will just placate you and give you the answers that it knows you want rather than challenging you.

Now, I do believe that you can guide chat GPT and these conversational AI's to actually challenge you too. So you could design, you could design it like that. You could actually, and this is maybe something for us to play around with Rico. You can actually maybe build a mentor. Right. You could build a mentor and say, Hey, you've got X number of years in the field.

You know, these are the things [00:30:00] that you studied. You've had challenges along the way, uh, be my mentor. Right. And, uh, and I'm certain that, uh, that Chad, GBT would, would probably, it's not going to be. Exactly as good as a real mentor, but I think that it would at least get you started. Right? So it, you know, it fill in some gaps.

And, uh, so that's something for us maybe to play around with,

Rico: Yeah, I think I, I think I've actually heard, uh, I think when chat GVT started coming around, there were people that were asking it, uh, you know, as part of the context that we're giving it or, you know, telling it what his position was, it was, they were, they would tell it play devil's advocate, right. They were, they were telling it to, you know, it's going to be contrary to whatever their request of it was, um, and had some success with that.

So I think, I think that would be a good thing. Like you said, it's getting the personal growth by. Meeting some kind of resistance or figure out that your way is not the best way. You know?

Mike: Yeah, exactly. Exactly. Um, and then, you know, I [00:31:00] think at healthcare, there are all kinds of. positives. Uh, you know, we're, we're seeing, uh, there was that article we did, uh, or well, we didn't do the article, but we covered the article, you know, chat GPT saved my dog. Um, you know, so that, that's one thing, uh, you know, uh, they're, they're, um, this was a couple of years ago, actually it's like five years ago now, but Google has done a lot with healthcare.

Uh, they've got a sector, a sector called deep mind AI. We've talked about it a bunch of times, but that, you know, a couple of things that I was noting when I was looking around today is they have, um, you know, AI that predicts acute kidney loss, uh, or lots of kidney function, uh, two days in advance of doctors, right?

Like two complete days. So if that's something that, you know, if there's even a suspicion. You know, you could use it for the diagnosis and get, you know, kidney function is very serious, like loss of kidney function. You're, you're going to die or like very quick, right? You're good. You could, you could, I wouldn't say you're going to, but, uh, but that's huge.

And then cancer [00:32:00] detection was another one. There are so many other ones too, but a couple were, you know, the kidney function, uh, and then, you know, cancer detection. Uh, so a deep model or a deep learning model developed by Google health. Which is owned by DeepMind, um, you know, in several, several medical centers, uh, they can identify breast cancer and screening, uh, mammograms, which is huge.

There are a bunch of videos on this too. Uh, we'll link at least one piece of content, uh, related to that, but I thought that was pretty interesting. 

AI Assisting the Disabled

Mike: And then you've got like, uh, you know, disabil folks with disabilities and really empowering them too, right? We've talked about, uh, Be My Eyes, and this is, you know, an AI application, uh, that, you know, see the world together.

And, you know, there is a human component to it as well, but really, they're trying to, you know, uh, empower and, um, you know, allow people with disabilities to make, to make better decisions, right? Because they have something guiding them or helping them. Uh, and then our ability. You know, this is really [00:33:00] exciting.

And again, we did a segment on this where, you know, they are, um, a lot, or are there empowering disabled folks and they're run, they're run and founded by disabled folks too, but they're empowering folks to, um, to get normal, you know, let's call them normal jobs, right. And jobs that they wouldn't be able to get otherwise.

And I would say that anything that can. Uh, allow a more level playing field, right for everyone, truly everyone, including people with disabilities, I think is massive, right? Like, I think that is just absolutely that, you know, it's just, it's just amazing because people want to, to, to, to contribute, they want to do something, right?

It's not just always for money. They want it, you know, purpose in the world, right? People want to have, you know, something that they can do that, um, you know, uh, you know, helps them grow and helps others, right? And, uh, and this is. You know, part of that, part of that solution, possibly,

Rico: And on that with the, you know, with the customer servicing and that, you know, [00:34:00] we have too many aggravations in life as it is. Correct. You know, like, so, so for the customer servicing, what do you have before you used to call a phone number, you would get an automated system and, and you know, the Uh, the advancements previously were what, uh, you could actually speak what you're looking for.

So if you want to talk to billing, say, you know, or tell me what the problem is, how many of us are yelling into the phone, you know, I said billing or I said this, or I said that, you know, uh, but now you

Mike: Well, not me. Not, I don't

Rico: All right. All

Mike: I'm always super calm with the

Rico: Yeah, sure. You are. Yeah. Right. Uh, and then, um, so now you can actually talk to the chat bot, kind of go back and forth and then, you know, figure out what the problem is and how to assist it.

And the same thing with trying to be more of an individual when you have some deficits, what better way than to have AI there to assist you, uh, understand your needs and, uh, you know, complete that task for you or, or get somebody to help you in a more efficient manner. So those are definitely the good sides [00:35:00] of, of both of those.

Mm hmm.

AI for Coding and Software Development

Mike: Yeah, no, absolutely. Uh, and, and then another area where I see it really, um, and this is, I would say could be its own whole show is, is code generation and code troubleshooting and whatnot. So this is like software development. Coding is really what I mean. Right. So I, you know, I I've built applications for many, many years and it isn't perfect.

Uh, you know, it's always getting better. It is definitely isn't perfect right now, but I see more and more the empowerment, enablement, and the expediency of a lot of these code generation, uh, tools, you know. Uh, leading to, especially in the future, I think, you know, a big error reduction in code quality, uh, improvement, right?

I think that's huge. Um, adaption and learning assistance, right? When you're learning a new coding language again, this is something that. Can be very difficult. Um, one of the things that GitHub Copilot does is you could put code in [00:36:00] or you could select code in your editor and you could say, explain this and that's really cool.

And then another thing that it'll do automatically is there's a button to translate the code to a different programming language, which is really, really neat because that's the thing that like, okay, I have a code sample that I know worked in my old thing, but I've got to translate it to a new thing.

I'm not sure exactly how to write that. Boom. You know, just go. And, uh, I was watching a video a couple of nights ago. We'll link it. Uh, it was, it was about a guy talking about really the future. Again, we could do a whole show on it, the future of, of coding in general and just coding, right? Which means again, typing the words, kind of doing some basic troubleshooting, whatever.

And, uh, and how computers are just so much better at that, uh, at that, that us, because if you, he went and he took like. Well, in a really old programming language, right, which looked archaic, and then he went through a couple examples in between [00:37:00] then and now and showed an example now of a hot language that everybody's

Rico: Right.

Mike: to, and he's like.

I can't read this. He's like, I've programmed for 40 years. He keeps like, I cannot read this right off the bat. I don't understand it right away. I will. I will. I could learn it, but I won't understand it. A computer language code done, literally language code done. And it might not get it perfect, but it's going to absolutely do better than you.

It's like a guarantee.

Rico: Oh yeah.

AI for Speed and Efficiency

Mike: Unless you already know that language, you know, um, and then speed and efficiency just overall from that. Right. Like again, you know, I, I hear a lot of people, Oh, it's got a hundred X and whatever, and maybe it doesn't a hundred X, but maybe at 10 Xs, maybe at two

Rico: Sure. Right.

Mike: And it depends on who the developer is too, right? Like a really good developer might actually get funny enough 10 X out of it because they know how to use it and they know how to get. Asked any issues [00:38:00] quickly, whereas, you know, somebody brand new, maybe, you know, they're going to two acts, but the reality is like a junior developer.

There used to be a junior developer that had one to two years, right. Where they're going through kind of just learning the craft. That process is going to be.

Rico: Way

Mike: I'd say closer to like six months if I had to think, because you need, you need to make a bunch of mistakes and it takes time to make

Rico: Right.

Mike: So the reality is like, you know, it's not going to come overnight. It's not going to go two years to two weeks, but it's going to probably go like two years to six months, eight months, something like that. All right.

Rico: And of course, Mike, as I always do, you know, taking it back to business, you know, when you're talking about that, I think about the business owner, you know, and, and their, their bottom line. Uh, think about that. You know, you're, you're having employees now that are, that are growing exponentially and shorter periods of time.

So you're paying them less time to learn, you know, so instead of two years to get up to speed, they're up to speed in six months and then what [00:39:00] are they producing? And the other thing too, on that, uh, is why there's such a big, uh, pull right now for prompt engineers, right? Uh, because they're looking for people who understand things, but also understand how to communicate with a system.

Uh, so. They get the answers they're looking for, because if you don't understand, again, if you don't understand quantum physics, how do you ask the system about quantum physics, it takes, you know, a little bit of knowledge about it to get you where you're going. Same thing with code development. If you don't understand the lingo or any of these language models.

You know, like to get into it, if you're not speaking the, uh, what you need to speak to get the proper output, it's not going to do you any good. And, and I think that's where people are going to fall short, especially with dipping their toe into it. Uh, they may just write it off and be highly skeptical, but the reality is just put a little bit of time and effort into it, drop a couple of good prompts in there and, uh, especially with these plugins, you know, the stuff that you can get out the other end is just absolutely amazing.

Mike: well, yeah, and you know, I will say this, um, you know, especially after using it now [00:40:00] heavily for a year, over a year now, uh, I'll say this, that, you know, You will, after probably three months, you'll start to realize that there are some limitations,

Rico: Oh, yes.

Mike: big pieces of text and multiple data sources and taking structured data, like what's in a database.

And trying to get that into what's called unstructured data, which is what these, uh, chatbots really focus, you know, like they're best with unstructured data, right? Um, and they could, I mean, a spreadsheet and whatnot. I mean, I'm not saying they can't do structured data. I'm just saying when you're building something with a lot of business logic and whatnot, really, you know, you want to.

Um, you want to have unstructured data, right? Or, or both, or both. And that's the thing, right? So, um, the reality is, you know, there's going to be, uh, a lot more, uh, advancement, I think, on, hey, here's how you build a complex application with conversational AI [00:41:00] as part of the solution and not the whole solution, right?

You know, there, I could build an app, heck, Rico, you built an app

Rico: That's right.

Mike: AI, right? Hit the API and, um, you built an app, right? But the thing is, once you go 10 steps past that, that's kind of like where you start to run into some like, oh, oh, here's where the problems are, right? This isn't, this isn't perfect either.

Rico: It's not perfect.

Mike: It's not perfect. It's not perfect. And then here's the last thing that we have in our, in our good column, right?

AI Creating Loneliness

Mike: Which also has a double edged sword, a sword, which is loneliness, right? So one of the cool things that, um, that, that Rico and I had found was, uh, they're actually having AI bots, like. These were physical robots go in and have conversations with the elderly folks in, in, um, you know, convalescent homes and whatnot.

And it was super cool. Like the robot would come in and entertain and like, it would talk to them and like, [00:42:00] have conversations. It could have personalized conversations. It can talk to a group and all of this to say that. You know, people's families aren't going to be able to be there 24 hours a day. Right.

And I will say that the staff is usually going to be, you know, like, I know that my mom's place is like sometimes scared. It's like, there's like one person for all these

Rico: Right. Especially nowadays. Every, everybody's short staff, every industry is short staffed and that's an industry that's been short staffed historically.

Mike: Correct. So, and that's just exacerbated these days. Right. So, you know, if you could have, um, you know, just, you know, actually not just keeping them company too. Like, I remember, um, my dad, when he was in the hospital, he actually had a robot that was like watching for certain things and like it could detect if there was like a certain set of problems and it would like call the nurse automatically.

Right.

Rico: think about that. Thermal imaging.

Mike: Yeah, exactly.

Rico: I mean, [00:43:00] it like it could count respirations, you know, the rising and falling of the chest. It could listen, you know. Yeah. The judge temperature, if you fall and you can't reach anything else, you know, to think about that for a second, instead of having a life alert around your neck, you simply call out or say command or something.

Mike: Yeah, yeah, yeah,

Rico: Yeah. I mean the, and the other thing too, you know, the, the cognitive decline that any of us would go through when isolated or even, uh, you know, I guess a minimal version of, uh, being isolated, but still, if it's you and one other person in another room and they're infirmed in some way that they can't speak and you're not getting a lot of interaction, but having them have a conversation and maybe even they could choose the character,

Mike: Right. Right.

Rico: the conversation.

Mike: Yeah, exactly.

Rico: benefits there.

Mike: You, you were talking about character AI, right? When we were talking about this earlier, right? Character AI will let you talk to a bunch of different characters, like ghost from call of duty was in there. And like,

Rico: Musk,[00:44:00] 

Mike: yeah, a bunch, a bunch of the characters from like Genshin impact.

So that's really cool. Like, and granted, like, this isn't going to be perfectly, it's not like actually having. A person there

Rico: actual conversation,

Mike: Right? I mean, honestly, a lot in the nursing homes, though. I think it might feel like that to those people. Right? Um, and then, you know, we think back to like the movie her right where, um, he had his own personalized AI and there was some pushback there.

I remember watching that movie and. Like there were some things that would, you know, and so I thought that was interesting. Now, granted like there on the loneliness side, the other edge of it, right. Is I feel like more and more like, think about the difference, um, that having a mobile phone made, right?

Like just being able to have a mobile phone, how much has that increased loneliness over the years? It's the answer is so, so fricking much. And,

Rico: at, look at public. Everybody has one in their hand. Nobody's

Mike: kind of what I'm

Rico: around them. Yeah. They're there walking around, looking at their phone,[00:45:00] 

Mike: Well, what are the, yeah.

Rico: the train, looking at their

Mike: sitting, sitting in a train, sitting in a concert, looking

Rico: hiking on a trail. I mean, I've seen, I've seen it all kinds of places where people are really recording, you know, I don't know if you ever seen that meme, but there's a perfect meme of it. And it's a later this older than, you know, the other people in the crowd at a concert. And she's the only person that's not holding a phone or anything has her arms folded in front of her and is watching the band for what it is in that moment.

And people don't do that

Mike: And exactly like, um, and, and while I was looking at something the other day and it was like, what's something, you know, from before high tech, you know, that high technology that, uh, that you really miss, right. And there were a ton of things that were

Rico: Oh, sure.

Mike: morning cartoons, having to watch the show at the same time of everybody

Rico: Delayed gratification.

Mike: delayed gratification. Right. But one of them was absolutely watching a concert in person without being blocked by a bunch of people with phones. And I said, that, that, that's a big one, right? At weddings, it's [00:46:00] like, how annoying,

Rico: Well, privacy alone, you know, you look at, you look at the direction nowadays as a result of technology and social media. And now, you know, now you can't make a comment that somebody disagrees with on a hot button topic without somebody saying, well, let me figure out who this person is and try to get them fired as a result of it because everything's so polarized and you

Mike: there's, and there's AI that'll do that. And well, not exactly that, but like, I was looking at something the other day where you could, you could put a picture in there's actually AI where you put a picture in and try and find that personal, like scour using AI, like the internet and all, and it's like, that's scary, right?

That's like a.

Rico: docs.

Mike: That's like, uh, stalker. ai or something, right? Like, for real. And, um, and so, uh, you know, something like that, that's, that's kind of like almost in the ugly area, you know what I mean? Like, so that's, that's, and we have ethical dilemmas in the ugly, uh, section, which is true, and we'll talk about some of those as well, but yeah.

This is, this [00:47:00] is a big issue. And, um, so, you know, moving on into, 

The Bad in AI

Mike: you know, our, our, the bad section. Right. So, um, one of the things that we have here is job displacement, right? Like clearly as AI advances that many tasks that were once performed by humans, uh, you know, are going to maybe automated, right? They, they aren't a hundred percent going to be automated, but like they may be automated.

And I think one of the biggest, forget jobs, like I think creatives. There's, there's really a big issue because we thought that it was gonna, you know, uh, take over all the menial manual jobs, like, you know, janitor or whatever. And by the way, there is a janitor AI that that's

Rico: I've seen bots

Mike: Yeah, exactly. Which is fine.

But here's, here's the thing, like, it didn't do that first. Like a lot of the, a lot of the stuff that we see, right, is like generative AI, which, you know, that could be art, that could be blog posts, that could be all these different things. And. So that people are like, well, you're taking the fun stuff away.

You know what I [00:48:00] mean? Like You know, it's really just displacement in general, and one of it being jobs. And, uh, and so, you know, it's going to have a significant impact on businesses that rely on certain industries for employment, right? So you're going to see a shift in. Folks needing to upskill. We've talked about that a million times, not a surprise, you know, but, uh, forward, I know one of their, their executives was saying, like, somebody asked them about AI and is it going to really displace a lot of jobs?

And he said, well, yeah, he thinks so, because people aren't going to upskill like they need to. Right? So this is a huge thing. And, uh, I think it's important for, you know, folks that, um, you know, pretty much in any sector and anywhere you're at in the workforce, figure out where this applies to you. It might not apply very much, but figure out where it applies to you.

Right.

Rico: right. And on that point, you know, like I told you, I had saw Ilan giving an interview recently and Ilan had said, you know, AI could make [00:49:00] it so that nobody has to have a job. In fact, only if you wanted to have a job, then you would have one. Uh, and, and think about the displacement. What if we, when we know people who work jobs, I mean, we've all had jobs that either we didn't care for that you kind of stuck doing.

But if you think about it, yeah, What if that displaces people and they find a better calling, a better way to spend their time, you know, something that's better for their personal growth. Those things are going to happen as well, because some of these mundane tasks or tasks that weren't done as efficiently as they could have been are now being, you know.

Replaced with AI bots, whatever you want to call them. Uh, but there's going to have to be people who do maintenance on those. There's going to have to be people who write the code for that. You know, there's going to have to be people who build the infrastructure. So those things are powered, you know, all of these things.

So it could be an advancement, uh, of humans overall as a race, because now we're, we have to move in and evolve. With technology. So

Mike: Right. Exactly.

Rico: I can see two sides to it.

Mike: [00:50:00] well, yeah. And, and I think you posted, uh, you sent me something earlier. It was like, um, if meme or something to that effect where it was like people in 17, whatever, you know, um, you know, uh, slates are, you know, I don't know why

Rico: Oh, yes. Yes. Yeah. The education thing.

Mike: um, writing on slates, blah, blah, blah, and

Rico: That's right.

Mike: and then like, you know, it was like to the

Rico: The ballpoint pen

Mike: a ballpoint pen will never replace the pencil, will never replace the pencil.

Oh, people are, are coddled with paper and you know, what's going to happen when, you know, that runs out, they're not going to be able to, uh, do what they need to do to communicate, right? Clearly, all of these things have, have went the way of the dodo and again, there will be certain things that you just like, I don't need to do that because it's okay.

It's covered. Right. I get to do, I get to do other things and I'm, I'm very much of the mindset that I'd rather do more interesting things and I'm not a mundane, like mundane stuff makes me insane actually. So

Rico: Right. [00:51:00] Tedious tasks, you know, and you're right. Like, uh, there's pushback at every level of that throughout history on education and the advancements in the educational system. So that's what we're going to see this time. People hate change. They're afraid of change.

Mike: right. Um, now another thing that, uh, that we've talked about multiple times, especially we talked about this in our, uh, responsible AI.

Bias and Discrimination in AI

Mike: This was episode 4 bias and discrimination, right? So, uh, what we know is that AI algorithms are only as good as the data that they're trained on again, going back.

to the data. Uh, if the data contains biases, the algorithm may contain those biases, right? Uh, it could lead to unfair treatment of individuals or groups. Uh, you know, and and some of these air like big decision things to like, like hiring lending, right? Like the implications of that are huge. It's not just, you know, that mid journey didn't wasn't super inclusive and what it printed out.

But but I'm not saying that's not a big problem. That's actually a really big problem because like what happens when you say, [00:52:00] um, you know, uh, imagine, which is the mid journey prompt command, imagine a beautiful woman. And it's, it's like four white ladies with super pale skin and,

Rico: One color, one type of dress,

Mike: Exactly. That is, that is super problematic. Um, and so, uh, you know, one of the things that we had seen, I think it might've been Harvard business review had said, um, you know, want less bias in your, uh, AI models, hire a more diverse and, you know, yeah, diverse, uh,

Rico: remember that article

Mike: Right. And, uh, so this is still, this is still really huge and then.

We've got privacy concerns too, right? This is yet another thing. This is a, so we're in, we're in the bad right now, right? So we have a job displacement or just creative displacement. We have bias and discrimination, privacy concerns, right? Um, Rico, we did a bit, uh, you had actually found the article where, what was it, Roblox and Call of Duty?

I know Call of Duty because I play it on a [00:53:00] regular

Rico: the censorship.

Mike: I think was the other one. Yeah, where, um, you know, it was, uh, you know, it's monitoring those voice lobbies and looking for something right. You know, hate speech or whatever. And again, we talked about, I get it. Like, I totally get what this is doing.

And I get that, like, straight up hate speech is, is a really big issue. And obviously, You, you know, you, you don't have enough people reporting people like, you know, the reality is there's a lot of false positives with those reports. AI can hear the set of words. And then the thing is though, like, how's it making the decision, right?

We don't know that and that goes back to an issue of transparency, right? So transparency in AI is like, Okay, you made a decision. Tell me the steps. Like, like, like my math teacher used to say, show your work. Show your work, right?

Rico: Yes. Yep.

Mike: Yes, and writers, right? Cite sources. Math, show your work.

This is a big deal.

Rico: that's big in [00:54:00] 2023. Even you, you look at it now. I mean, you talk to anybody and I don't care what the topic is. You talk to somebody nowadays and they say, Oh, I, you know, I saw on the news or I heard. You know, which are words that now is like, well, what's your source sites? Your source, you know, that's, that's become a big thing over the past two or three years.

It's going to be that way with AI. And that's what makes me skeptical too. And we've talked about this before. I've made reference to it is who's on the backend of that AI. What are they telling it? You know, which, which way is it leaning? Which, whether it's culturally education wise, politically, religiously, which way is it leaning and which way is it going to direct me, how is it directing my children?

How's it directing culture? You know,

Mike: Which is, which is why, which is why what I'll do with, um, with, with my different prompts, especially if I think there could be a bias, I will actually throw it through three different, uh, conversational AI's that I know don't share the same back end. Right? So, [00:55:00] so just to let you know, folks being an open AI, which is chat, GBT and perplexity, they all share.

Actually perplexity gives you options, which is really cool. That might actually, it might be a little a bite on that, an AI bite, but, um, they share the same backend, right? So Claude is, is another tool, uh, made by a company named Anthropic, and they actually have their own like engines and backends. So it's not using, uh, the G pt, you know, the, the, the open AI GPT models, right?

And uh, and then Bard is using its own thing. It's not really. That great though, honestly, I haven't even played with Barton in a

Rico: I've never got it to work. I honestly, I've tried it like two or three times. It's never worked for me. So I haven't been back.

Mike: yeah, so, so, but the reality is, yeah, the reality is that, uh, you know, I like to kind of get an idea of what things are, you know, like, what's different between the three answers if I ask it like the same question and I, and I do take into account.

That all these tools are non deterministic, [00:56:00] which means they're not going to give you like, you can, I could hit go on that. I could open three chats, the exact same starting message, and I'm going to get back probably, I mean, at least slight differences in the answers. Right. And the same engine. So open AI, let's say chat GPT.

And, uh, and so, you know, that's, um. You know, that's a, that's a big deal, right? Uh, but, you know, one of the things that, things that I wanted to talk about with privacy concerns is, you know, AI can analyze huge amounts of data, right? We talked about Roblox, we talked about, you know, Call of Duty and whatnot, and, uh, monitoring these, uh, you know, monitoring these You know, spaces, right?

And, you know, there's more and more, uh, opportunity then for our data to be misused, breached, uh, could lead to personal identification being exposed. I mean, I feel like, um, identity theft, identity, identity theft is not a joke, Rico, that's all I know, but, uh, I think it's more and more, uh, omnipresent in the world that we live in this, [00:57:00] you know, these days, right?

Because, uh, it's, it's, you know, a lot simpler. For people to get a hold of some of this information or socially engineer, right. With, you know, with deep fake voices, um, you know,

Rico: That's going to be a problem.

Mike: that's going to be a real problem

Rico: already seeing it, you know, you're hearing about it. Uh, well, we talked earlier about the Scarlett Johansson, right.

Mike: Right,

Rico: the lawsuit because of, uh, her likeness and a voice being cloned. Um, and you know, you and I have done our own voices and cloned them. Uh, you sent me a few bites with it.

Was that yesterday? Or I think it was yesterday we were going over it. Uh, and. It sounds just like you and the change in the sentiment and you, you know, the, the, where you put the emphasis on things, whether it's breathy or not, again, we've talked about that before when we covered 11 labs on how stable you can make it, it's, it's absolutely impressive.

And those could be used, there's always going to be nefarious people out there doing bad things. Um. Bad actors and stuff. And that's, [00:58:00] that's going to be in the bad, but yeah, clone voices and deep fakes

The Ugly in AI

Mike: All right. So yeah, uh, so, you know, really, uh, you know, we're going to jump right into the ugly, right? We're talking about deep fakes Rico. And, uh, and I'd say that that's one of the things that I would call ugly, whether it's, you know, deep, big voice or deep, uh, deep, big video. Uh, you know, it's a concern, right?

Because you could get a call from, uh, you know, your daughter asking you for money. And it's, it's not actually your daughter, right?

Rico: And we've seen those. And what's interesting is, uh, I've seen, uh, over the past week or so that they, they're actually trading in now to these models that you cannot generate, say generative art, for instance. You cannot, uh, generate, uh, political figures in compromising positions. Uh, and there was, uh, one gentleman who had a whole channel dedicated just to that, and of course he was doing it to be funny or [00:59:00] to test the boundaries or whatever, but, uh, yeah, those are, those are things that are.

Uh, they're going to have to put things in place to stop because people are so quick to believe things nowadays and not, uh, discern and without having proper discernment. You know, you see something on tech talk at you and I keep talking about tech talk the past week. I've seen two or three tech talks that are, uh, crypto giveaways.

Uh, Warren Buffett is giving away crypto. Uh, and again, uh, you and I have, you know, kind of a trained eye at this point because we've toyed with these systems. I can see how an average person would fall for it. They aren't, it's not great AI, but I can't tell you how many times recently I've heard people say, well, I think that's AI.

No, that's actually that person speaking. That's not AI, but it was a message they didn't know. Uh, and then the ones that are, you know, deep fakes.

Mike: Yeah, is that AI? Is that AI? Well, my dad runs my dad runs a art on [01:00:00] art Facebook group and uh, And he's like and he does art too. He paints and he's like, you know what? I showed him the AI tools and he liked it. He thought it was cool But he's like I I won't I can't post any of that on my channel because what'll happen?

Is, um, you know, people ask if everything's AI, right, which is, you know, you, you put that seat of doubt in and, uh, and that, that, you know, that's gonna, you know, it's gonna always be a question, right, where he just doesn't want to hear it. Right. Um, another, another deep fake that, you know, came up recently, I guess this is a little while ago now is they had that Pentagon hoax, right?

Like where the Pentagon was on fire. It was posted on Twitter. Um, and, uh, and you know, people thought it was real for a little bit. I mean, at least for a few hours, right? Like news, other news outlets were actually grabbing it and like putting it in their thing. Like, you know, it's like without fact checking again, critical thinking folks always.

And, uh, and so that, you know, that, that's a big issue. Right.

Rico: Oh, on that, just, just to that point, Mike, uh, you know, with [01:01:00] all the, the, the wars and things happening or the rumors of wars, I guess it's happening currently, there was the, um, the video of the vehicle burning in the intersection. You saw that one that was made with unreal engine.

Mike: Oh, yeah.

Rico: which has been used for, you know, for video game creation and stuff.

Uh, and there was another one that was a AI generated baby, uh, that was in the Middle East, um, in a compromising position, you know, in harm's way. Uh, and there were a lot of people that fell for that and the news ran with it. I took one look at it and was like, that's AI. It's obviously AI.

Mike: I honestly didn't, I didn't catch

Rico: You haven't seen that one.

Yeah. Uh, it looks like a baby taken from rubble, I guess is the easiest way to describe it without getting too, uh, graphic. Um, but, uh, just looking at it, if you've seen generative art at all, it's clearly AI, but it had a very powerful impact and those things have a negative response as they should, uh, to somebody being harmed, uh, that's in the, that's in the bad category for, for generative art being used misused, I

Mike: Yeah. Well, the, [01:02:00] the ugly, right? I mean, it's really the

Rico: the ugly. That's truly the ugly.

Mike: of those things that like nobody, like there's not anybody who even the people are making it like, no, they're, they're objectively wrong. Right. Like, it's

Rico: Oh, yes. They know it's going to get a response.

Ethical Dilemmas in AI

Mike: Yeah, exactly. Exactly. Um, and then you've got ethical dilemmas, right?

We mentioned that a little bit earlier, but you know, uh, it raises various ethical questions such as responsibility for AI's actions, right? That that's a big one, right? Who's responsible if AI, you know, tells you to do something or let's say, let's take it a step further. Let's say AI is implemented in a robot and AI Makes a bad decision.

Like I had talked about something recently where they were just ran a simulation and, um, you know, it was a, it was a jet and a pilot or it was an AI, it was an AI jet and, um, but it was piloted and the only way that it could like accomplish the objective it needed to accomplish was to kill the person like that was manning it, right.

And, uh, [01:03:00] and so, I mean, that, that's huge, right? There's something called alignment, which really is like, Hey, is AI needs to be aligned with the goals of humanity. And we talked about the other thing, which was very hypothetical where, you know, okay, you give AI an objective, like, um, you know, clean all of the oceans, right?

Like complete, make them completely pure. And AI is like, all right, cool. I can do that with, Oh, it has to be within a year. I could do that within a year. But then you real, like AI just does it and then you realize that to do it and needed to take like, you know,

Rico: Get rid of humans.

Mike: 20, no, 25 percent of the world's oxygen, which then we're, you know, we're slowly dying, right?

Like over the next, you know, however many years. So these are things that, uh. You know, are, are going to be more and more programmed in AI, which you'll, you will see a bit of, um, let's just say it'll, it'll feel a little bit neutered to some degree in the same breath. Um, you know,, [01:04:00] there are, there are lines, right?

And so there, there, there are whole frameworks around building in such a way where you don't. Lower the ability so much of the AI, but in the same breath, you're, you're not doing things that are blatantly, and there's a wide spectrum of those. And, um, and so, you know, things that are just blatantly, blatantly, obviously not the right thing.

So they'd be in this ugly category. Right. And, uh, and I think that open AI has done a pretty good job with that. I will say, like we talked about, it can be a little preachy, right? Like you weren't able to, or. All right, you weren't even you probably were using dolly or something but whatever you were using like it

Rico: That was that. Yeah, that was dolly three, uh, on the, on the image generation. Uh, but, but the other thing that we were talking about previously was, was chat GPT. Uh, and again, I, I, there was no intention to harm anybody. It was just the, the particular statement mentioned the crime and therefore chat GPT. And all I wanted to do is, you know, [01:05:00] a series of looking for patterns that, you know, I don't look for to determine deception and it wouldn't do it for me.

It wouldn't even entertain it. And I went a step further to try to explain it to it. And I've heard on other, um, There's other tech folks that I've been talking about. There's ways to train that out of the model itself by using go arounds. But that, that's what they're, you know, the people who are building these things are trying to prevent is that people can't do work arounds to get the information that they need.

Um, and on, on the ethical portion, you were talking about, um, who gets to determine what's ethical and what's not. You know, that's, that's the big question

Mike: Yeah. The people that are making the rules, right. Or are the

Rico: who makes the rules.

Mike: again, bias, right? Like it, they get back and it's like, all right, the person who's making the rules is the one who's saying what's right and wrong. And that's why you need that group of, of folks to be extremely diverse.

Right. And at some point, um, I I'm certain we'll see this. If we're going to see personalized day, I will [01:06:00] definitely see this, which is like regional AI, right? Like, so, you know.

Rico: We talked about that.

Mike: because we don't want to, we don't just want to whitewash the whole world and give everybody the exact same answer, right?

We have culture. We have all these different things that should be celebrated. Right. And, uh, you know, people need to be able to maintain their individuality as well as You know, their, their, their culture, the regions. I mean, this is beautiful stuff, right? Like we want to maintain that rather than having, um, you know, everybody, you know, it's all, it's all the same, it's all the same, right?

So the beliefs are all the same and, you know, the, the world has gotten into trouble, you know, with people who, who wanted that, right? And, uh, so we, so, so we. Yeah, exactly. So without going any deeper with that, like we want to avoid that, right? Let's, let's avoid that. Right. And, uh, so yeah, so, uh, and then we have existential risks too, right?

Like, um, this is, this, it's funny that, uh, Elon Musk named, [01:07:00] uh, his AI XAI because a lot of people, uh, when they think of XAI, they think of existential AI, which are like the existential risks of AI. Right. So, um, you know, long term impact on humanity, right? Including the potential for AI to surpass human intelligence and control.

Uh, and it's really, you know, to me, people always think of the Terminator and these different things. To me, I always think that it's going to be bad actors with AI. That will destroy the world. Well, before AI becomes, you know, it's just making the decision to, uh, you know, because you think about how much training, how much negative training it would need to, and how much you'd have to like completely remove the guardrails, right.

Like, and actually probably guide it. And, uh, again, like I said, influences of, you know, uh, you know, malintent, uh, malintended folks. Right. So, um. You know, that's my take on, on that. I mean, [01:08:00] I do think that there are definitely some existential risks when we're going to see those play out. I think it's probably going to be a while and I don't see a path from A to whatever it ends up being.

Rico: Right. And it's existential.

Mike: It acts so, um, so that's it. But, you know, and I will say some experts, you know, speaking of Elon again, um, you know, and Stephen Hawking, right? They've, they've worn against this, right? They are concerned and they want to make sure, uh, that, you know, this is handled in a responsible way.

And, you know, there's a point where. You know, it's kind of funny because six months ago, uh, and this isn't the bash, bash Elon, but six months ago, Elon was talking about how we need to put a pause. on AI. And I literally said, I said, yeah, cause you got something cooking. Right. And now, now, hang on. Now, one of the things he says about this XAI, it will be the most powerful AI.[01:09:00] 

Rico: I remember you having, we had this conversation.

Mike: didn't even like, but it's funny. It's throwback to like May or June or whatever we would, you know, when we first started out, we actually had that conversation and one, one quick note too, about XAI, like Elon's XAI is that, um, you know, yeah, it's going to be, you know, best in class, right.

Which is, you know, compute class and stuff like that. It's going to be in a different compute class than like OpenAI's chat GPT and whatnot. Like it is what it is. Um, just like LLAMA or any of these other ones. Right. So, uh, take that with a bit of a grain of salt. I, I think that, uh, Ian or Elon has a lofty goals and I love him for it.

But the reality is, you know, like, um, it's yet to be seen. We'll see. I got to imagine it'll be better than Bard though.

Rico: Oh yeah. It's going to be better than Bard hands down. I also, on that, I find it funny that, you know, we're okay to put a Neuralink in a mess with people's physiology, you know, put something in your brain, uh, you know, [01:10:00] he's got that, right. That

Mike: I, you know, what's, you know, it's funny. I. Yeah, I actually almost threw that in as a news story if it would have the time, but well, I'll just say about this. That's another one to me that could go good, bad, ugly because I think that the at least the spoken intention is really good with Neuralink. I love the idea behind it just to give folks an idea of what it is.

Um, it is an implant or implants in your brain. But it allows people, it allows these, the, the, the neurological links to be, uh, bridged, whereas they might not be right. So blindness and different, different ailments like

Rico: Paralysis. Yep.

Mike: is another one allowing people with these ailments to be able to function as, you know, like repair those, or at least let's not say repair.

Let's just say like bridge the gap between their, their neural links. Like that is, uh, their neural

Rico: part of it is great.

Mike: that's so great. But think about something's in your

Rico: Hackable networks.[01:11:00] 

Mike: networks.

Rico: That's what, that's where my brain

Mike: fricking crazy is that? Right? Like, so that could be good, bad, ugly. It could be a whole thing.

Right. And, uh, that, that's a good one, honestly, to, to kind of end on here. I think those are the, unless Rico, do you have any more that you had for any of the, you know, good, bad, ugly. I think. I covered all mine.

Conclusion 

Rico: No, I just think, uh, that really it's in the eye of the beholder, right. And that many of these things are double edged swords, you know, it's too, it's too early to tell. We try to cover things and bring it to you as it's happening. But, uh, there are many of the things that do have two sides to it. And again, uh, I think it's going to come down to how it's applied and who makes the rules.

So

Mike: Yeah, absolutely. And, uh, and folks, thanks for joining us tonight. Uh, you know, we'd love to hear your feedback on this episode. I think there's so much room for conversation with, uh, you know, with this subject and I'm sure that there were things that we missed, right? Because I feel like I had. You know, [01:12:00] a hundred more things that like

Rico: sure.

Mike: couple of weeks that went through my mind.

I'm sure I missed some of those. Right. So, um, please, you know, reply in the comments and, if you like what we're doing here, give us a, like, give us a subscribe. We're putting our content at least, uh, an episode a month. Uh, you know, we've been pretty consistent with that and, uh, we, we'd love it to have you join us on this journey and, uh, and delve into AI.

Rico: See you back in the lab soon, folks.

Mike: Thanks. All right, folks. So see you in the lab again soon with Mike and I, and we are artificial antics. Thank you.