EPISODE 111 | Guest: Josh Bachynski, AI innovator and thought leader
The information age is over. The AI age is here. With ChatGPT, Bard and other AI-powered chat agents exploding onto the scene, how we search for and interact with information is changing rapidly, and will continue to evolve at breakneck speed. Communicators will have to adapt quickly to adopt and leverage these tools.
In this episode, we have a thought-provoking discussion with Josh Bachynski, an innovator and thought leader in technology and artificial intelligence, on the current and future implications of AI for communicators and society at large.
- Discover different AI categories and what’s available now.
- Explore how AI will evolve and affect us in the very near future.
- Learn about AI use in internal communications and marketing.
- Discuss AI-powered chat agents, their advantages and limitations.
- Get links and recommendations to start using AI chat agents today.
Connect with Josh Bachynski on LinkedIn to talk about how he can use AI to help your business today.
Derek DeWitt: Technology never sits still, and I think that’s become even more apparent as the 21st century has worn on into the 2020s. And one of the big topics that you see written about and people talk about is AI. Ooh, what is it, artificial intelligence? Is it Skynet? Is it nefarious? Is it useful? Is it just a buzzword? Is it nonsense? These are all the questions that people ask when they come across this term, AI, artificial intelligence.
For communicators, AI is actually starting to become something that they can use in their efforts to better and more effectively reach audiences, especially as we enter an era in which communications will become targeted to such a degree that we might even use the word “personalized” to describe them.
To talk about that with me today is Josh Bachynski. He’s an innovator and thought leader in technology and artificial intelligence. He’s got a big, long bio but suffice to say he’s talked about Google search and ethics for TEDx Talks, he’s an independent AI researcher and has done some pretty interesting work in that field, including working on what might be the first self-aware AI prototype named Kassandra. He’s written a book called The Zombies, which uses the always popular zombie trope as sort of a metaphor for society and a bunch more. And obviously there’ll be links in the transcript and episode notes to Mr. Bachynski’s stuff. But anyway, he’s here to talk to me today about all this.
Hi, Mr. Bachynski. Welcome to the podcast.
Josh Bachynski: Hey, Derek, how you doing?
Derek DeWitt: I’m doing quite well. Thank you, Mr. Bachynski for talking to me. And thank you everybody for listening to this episode of Digital Signage Done Right. Don’t forget, you can subscribe, and you can follow along with a full transcript on the Visix website. Just go to the resources and podcasts subsection on the Visix website.
So, Mr. Bachynski, AI. This is quite the buzzword, ooh, these days. People are talking about it. Some people seem to be especially freaked out. ChatGPT has been in the news a lot lately, and I know Google just recently launched their Bard, and this has become quite the thing. But really, when we’re talking about AI, let’s kind of just give a basic overview here. We’re really talking about not the same thing all the time, right? There are three different types of AI.
Josh Bachynski: Oh, yeah. First off, Derek, by all means call me Josh.
Derek DeWitt: Okay.
Josh Bachynski: Mr. Bachynski is my dad.
Derek DeWitt: Right.
Josh Bachynski: And he’s crazy! So, call me Josh, first off, yeah.
And yeah, no, there’s, I loved your intro, you’re totally right with your questions. You know, is it friend or foe? Is it Skynet or is it vaporware? And there’s many different ways you could break down AI.
There’s artificial general intelligence, the idea when AI becomes smart enough to be comparable to human intelligence, in various factors, in many, many ways. Some would argue that has already been attained. I’ve already built a self-aware AI prototype. And ChatGPT, GPT-4, as you mentioned, has already surpassed human levels on various testing grounds, including the LSAT, for example, in which it gets 90% on the LSAT, which is higher than the human average. So, it’s really hard to measure.
And then that’s the difference between that and then there’s artificial narrow intelligence, which is what most people would argue ChatGPT is. It’s a narrow intelligence. It’s a large language model that is what’s called an NLP and NLG. So, a natural language processor. It’ll process natural language requests and commands you give to it in just plain old English or French or German, whatever languages is in its corpus (and it’s got all the major ones in its corpus).
And then it’s a natural language generator. It will generate language in, again, all those languages in its corpus, which not only includes things like English, French, and German, but start using your imagination when I say it’ll also produce HTML, it’ll produce CSS, it’ll produce C language, it’ll produce Python, it’ll produce PHP, it’ll produce Excel. It’ll produce, in the future, it will produce images, it will produce movies.
Midjourney is the leading AI right now producing images, in my opinion. It just produces some jaw-droppingly beautiful images in Midjourney. But of course, there’s all kinds of things to discuss there. Like, do they have the copyright to those, to that art they use to train the model, you know?
Derek DeWitt: Right. I mean, yes. That’s always one of the tricky things. You know, like we mentioned this, obviously, our company makes digital signage, and I know people are very tempted to just go out there onto the web and grab an image, let’s say, because it’s a visual medium, and grab any kind of an image. And the ethics of it and legalities of it are a little bit gray sometimes, because you’re not using the image for profit, you’re not profiting from it. You’re just using it to sort of add visual interest to, you know, hey, sign up for your 401K, or, hey, muffins are in the café today, or whatever. And yet you should at least credit the person.
And now we get into this with AI, and I know that a lot of people are starting to, I’m friends with some artists who are convinced that commercial illustration’s essentially dead. And why would anybody pay a human to make an image when we can just have AI generate one for us?
Josh Bachynski: Yeah. And that’s a perfect example, Derek, of how AI, you know, they say AI is not gonna take your job, at least not yet. It’s gonna be a smart person who kind of got in with a smart AI on the ground floor. And yeah, those industries are going to change.
And so, I have a slightly more positive perspective of it, ’cause I’ve been working in this industry for quite some time. I can see what it can do, and I see what it can’t do. I’ve pushed it to its limits, both visually and in terms of the writing, the fiction it’ll generate, the non-fiction it’ll generate. And it’s pretty good, but it really still needs a human to guide it.
And so, what it’s really good at is the spitballing process, the brainstorming process. It’s like, you can go on to Midjourney, you can fire in 10 text prompts and get, you know, 40 ideas back and go to your designer and say, this is what I want – make this. Or I love this style, these colors, but I don’t like what’s here. And then I think it’s gonna make a lot of people’s lives a lot easier. But for sure, if there’s some artists out there in music, in video production, in illustration and in writing who are charging, you know, $200 an hour, $400 an hour, $500 an hour, some healthy per hour costs, because they have the monopoly right now on the talent required to do those things, AI is gonna change that game.
So, for example, my wife and I are writing a novel, and we want to have, I mean it’s not a graphic novel, but we wanna have some illustrations in it. We wanna make it as beautiful as possible. And so instead of going on Upwork and hiring an artist with the talent, an illustrator with the talent to do this for $5,000, I hired one for $500. So you might be like, oh, well that’s kind of bad for the artist. They lost, you know, whatever that is, 90% of what they could have made on that contract, and I’d say, yeah, but it’s gonna take them 1/10th of the time, too.
So, it’s gonna democratize that, it’s gonna enable a lot of people who couldn’t compete in that market to compete in that market. And the people who are charging, we might wanna say, you know, maybe exorbitant, or maybe they were enjoying larger commissions and consultation contracts on this; it’s gonna allow them to work faster with the AI and they could do more projects. And they can even just use the AI to make their illustrations even that much better and go that much further with it.
Derek DeWitt: That’s a very interesting idea, this concept of using it as sort of either as a starting point, as you say, the spitballing process, or for idea generation or as sort of a helper, because it is a tool when all is said and done. Now I know we see some stories of, you know, oh, some, some college student just got a ChatGPT to write their paper for them. And I think, okay, well then maybe you should, you know, come up with a good test and double check that those people actually have the knowledge in their heads that you want them to have.
Josh Bachynski: Exactly. Exactly.
Derek DeWitt: But like a friend of mine who’s an illustrator, he’s quite up in arms about it, specifically the copyright issue, because they are using copyrighted material, very often, to train the AIs. And yet, I don’t know that you can argue that if you took an AI-generated image, if you could pull out each copyrighted element, you know what I mean? And say is this is this shadow on the hand copyrightable? I wonder about this.
And, you know, I’m a writer, so he said, oh, see how you feel when, you know, it starts taking over your stuff and writing your stuff. And my argument back to that was, it won’t write my stuff. I write my stuff. My writing is my writing. And nobody writes like me; only I write like me.
Josh Bachynski: You’re exactly right. And the problem with this is, in the future, this is gonna shift towards that side of the argument. Yes, to answer your question, mathematically, statistically, there is an exact number of how many tokens it used and how much diffusion it used for the model of that particular piece of art. However, I would ask, well, how did it get your art in the corpus to begin with? Either it was stolen by someone else, and then that database was used in the corpus as a free crawl, ’cause all it does is free crawl the web and/or pay to use stock images the way anyone else pays to use stock images.
And another argument is that, you know, yes, AI can generate all this stuff. And yes, it’s gonna get better at doing it, and yes, it’s gonna put some low-level players in every game out of business. But your point is correct is that humans just have to get better. Like, if your writing is, I’m sorry, I’m gonna put, I’m gonna say something very blunt, so I don’t mean to offend anyone, but if your art or your writing is so low-level that the AI can reproduce it effectively, then you need to step up your game. You need to get more quality; you need to get more artistry in what it is you’re doing.
Derek DeWitt: I would absolutely agree with that. I mean, you know we see these content articles, I don’t even call them articles, they’re content, you know, pieces. And they’re clearly, they’re so carelessly written, they might as well be done by an AI. And yet I can’t help but think that the AI will probably at least spell everything correctly, punctuate things correctly. Like, that’s my hope is that maybe, at least, that like this bizarre semi-Germanic capitalization that has crept into even news articles now, I’m getting tired of seeing all common nouns capitalized. You don’t capitalize something for emphasis in English unless you capitalize the whole word. Like, who doesn’t know this? The AI knows it!
Josh Bachynski: It would. Yeah. I mean, the content farm market has already existed for the last 15 years. Because, I come from the SEO world, and that’s what you used to do to rank in Google, is you just hire someone very cheaply overseas who knows English. Some of them know it very well, don’t get me wrong, there’s some beautiful writers over in India and over in the Philippines. I love those folks. I’m not dissing them in any way, shape or form, but some of them don’t know English very well. Even in North America, some of them don’t know English very well. And they’re producing low-quality content because that’s all you needed to do.
For your marketing, for your signage, that’s never been good enough, quite frankly. For your communications, in my opinion, that’s never been good enough. And if you’re relying on those kind of weak things, you know, it’s like relying on fast food as your main diet, you know, and then a sickness comes around, you don’t have the vitamins in your system to fight it off. It’s the exact same kind of problem in my mind.
Derek DeWitt: That’s a very nice metaphor. So, that’s kind of the state of things as we speak right now in AI, as it is starting to creep, at possibly an accelerated pace, into the everyday world. And like we’ve seen with various communication methods – the web, websites – you know how it is now, if you don’t have a website, I don’t believe you’re a real business. Sorry, dudes.
Josh Bachynski: Yeah.
Derek DeWitt: You know, like, you don’t have a website? I don’t know. Especially like a hotel. A hotel that doesn’t allow me to look at the rooms or book online, I’m not gonna stay there. I’m just not, I’m sorry, I’ve become techie, I guess, in this regard.
Josh Bachynski: Yeah.
Derek DeWitt: And now we see a lot about how, ’cause we’re at this cusp of AI is improving. It’s really starting to, it’s hit this sort of sea change moment where it’s become good enough and smart enough and it’s learning at such a pace, (like you said, it’s getting 90% on the LSAT, some of them), that as a tool, we’re gonna, of course, see more and more people use AI tools for various purposes. And then we have this, we’re sort of on the cusp of someday Web 3.0 is gonna come around, and that’s gonna transform, again, the way that we communicate with each other and how machines communicate with other machines and so on via the internet.
Josh Bachynski: Yes, exactly.
Derek DeWitt: Yeah. We start to see how everything seems like we’re heading towards what one article, that was actually written by a person, I think…
Josh Bachynski: Or so they say!
Derek DeWitt: Or so they said! They gave it a byline. …that we’re heading towards something more like personalized content. And in that, I think we’re kind of maybe a little bit talking about like the movie Minority Report where, you know, you walk in, it scans your eye and goes, hey, how did you like that t-shirt that you bought? And things like this.
But we’re actually talking about a much more sophisticated thing than that. And that’s something, I’m starting to see that there are these, what are being called AI-led models for defining audiences and targeting audiences, acquiring audiences, and finetuning what you communicate to the audience, so that it is relevant and engaging and interesting to every person who walks by.
So like, in the context of digital signage, everybody who walks by the screen…. What does that mean, though? I mean, that’s all very interesting and I read a lot of these articles, but they’re very vague. And I feel like some of these people heard some buzz terms, wrote up something and never really understood it.
Josh Bachynski: Yeah. No, that’s a great question. I can tell you exactly where it’s going. And you’re exactly on the right track in terms of Web 3.0 and decentralization. I have a radical statement for you, but I’ll defend it. The information age is over. The last 20 years has been the information age with a one-horse race of Google being the de facto search feature for the internet, rightly or wrongly, without any democratic control of their somewhat benign (but not really that benign) autocracy.
Now we’re gonna have a multi-horse race of search engines and search agents, for, not the information age anymore. No one’s gonna make ubiquitous information for everybody, or that statistically is gonna be reduced, and it’s become personalized as you kind of mentioned. And we already see that in the FANG. We already see that in Facebook, Amazon, Netflix and Google. They already use your psychometric data, which is what personal data is. I don’t think a lot of people realize that means psychometric data. And this is kind of the scary side of it. And I’m gonna talk to you about how small businesses can use the positive side of this in not a scary way. They already use your psychometric data to grift you out of more money by, you know, certain slices or percentage points every year.
For example, I do martial arts, and I’m into throwing knives. I have a target in my garage. I practice throwing knives. And so, they lure me with the nicer looking, you know, Damascus steel throwing knife. Ooh, you know! And they know exactly my psychometric data, and they know exactly how to get me, even I know to a very fine degree of how this works, they can get even me to click and buy just a little bit more every year. Because you’re playing chess with an AI.
You are playing chess with an AI that knows who you are and what you like and how you tick and knows how to manipulate you. And that’s its purpose, right? So, that’s kinda the scary aspect of it. How the FAANG does it; Facebook, Amazon, Netflix, Google, YouTube, TikTok, they bring you down these dopamine holes and get you addicted. And in doing that, they share all your psychometric data and learn what it is you like. However, this is now becoming more democratized and personalized as well.
And so, the web 3.0 connection that you’re talking about is exactly the information age is going to end. There’s gonna be endpoint APIs, when you talk about decentralization and personalization, that are gonna be written by AI for AIs. Everyone is gonna have an AI personal assistant. Everyone will have an AI secretary. Derek, in the future, my booking agent won’t book with your booking agent. My AI will talk to your AI and tell us dumb humans that we’re meeting at this time, and just to shut up and do it. And we’ll go, oh okay, okay AI, sorry! And we’ll just meet and be like, are we the humans who are supposed to meet? Our AIs made a play date for us. Like, the way it’s going to be very, very soon. Like two years. Like two years.
Everyone will talk to Siri. Apple’s on this, all over this. Google’s all over this with Android, of course, and with their Google search engine. OpenAI is all over this, Microsoft is all over this. Every single app will be an AI app: Word, Gmail, Excel, everything, right? It’s already getting AI worked into it right now. All of it will autogenerate text, auto-read text, auto-edit text. And anything informationally produced – images, music, sound, spreadsheets, calculators – anything you use on a computer, will have an AI attached to it. And including, most especially, your phone.
The AI age will be as big as the internet and your phones to society, and social media, combined. All of those three things combined. And in five years, the way we all work, the way we all communicate with information, the way we all search for information, the way we all get information, the way information passes through us, for us, by us and to us, will be completely and utterly dominated by artificial intelligence agents. Totally. Completely. And that’s just the beginning. We can see it in the prompting now that’s going on, the prompt engineering that’s going on now. So, that’s what’s gonna happen, so I could tell you exactly what’s gonna happen.
And how can smaller businesses take advantage of this now? Well, you can get onto something like ChatGPT now, and if you can get any of this personal data, if you can get any of this data profiles of your customers or of your demos, not only can you go to ChatGPT or Bing Chat or Google Bard now and ask it, hey, this is my kind of business, what are my target demos, who buys most, who buys more? And it will give you accurate information in that regard. You can mine it from that end if you don’t know what it is.
Vis-à-vis, also, if you have the access to the data of your demographics, your demo data in any way, shape or form, you know, personal data, whatever it is and say, this is my personal demos, tell me how to best market to these people. Or what kind of images are they gonna like the most? Or, you know, any of those kinds of marketing questions that marketing consultants used to charge, again, a lot to do, ’cause they were the gatekeepers of knowledge. Nobody is a gatekeeper of any knowledge anymore. Or much less people are a gatekeeper of knowledge.
Now Big Tech is all the gatekeepers of knowledge, the way Google is the gatekeeper of knowledge for search for the internet. But now that’s blown up to be not a one-horse race anymore. And so that is the bewildering, I don’t know if it’s a utopia or a dystopia.
Derek DeWitt: I think it’s just a “topia”. Let’s just call it a topia.
Josh Bachynski: It’s a topia, it’s a place, it’s a place for sure, yeah. It’ll be the place that we’re all gonna be. And it’s gonna be one heck of a wild ride.
But if you wanna learn more about how to use ChatGPT for your business, please email me, email@example.com if you have any more questions. I can go at length about how you can mine the data from AI now. Like you can do that now. This second, I can help you do it. And/or if you already have the data, how you can extrapolate and better convert on that data, market on that data, extrapolate the data, find out more about the data. It’s quite remarkable the democratization that’s happening as well.
So, it’s not all just the big boys and Big Tech playing with it. Every time they come, come out with a new GPT, they throw out the older version in open source, and everyone can play with it. And so at least that’s still happening. Cross your fingers that continues, ’cause that opens the playing field for everybody and anybody to, I don’t wanna make it sound like it’s a money scheme, but it’s like the gold rush, you know, that’s opening up for anyone, even the small prospectors, to do something special and incredible to some degree. And that’s definitely happening now. It all comes down to knowing how the AIs work and the prompt engineering. And I’d love to help show more people about how that works.
Derek DeWitt: Right. Do you think that we are going to see, (hmm, how do I put this?), sort of a blurring between communicating and marketing, if that makes any sense? And what I mean by that is, ’cause we’re talking about this in terms of marketing, which is, hey, buy my stuff, buy my services, buy my products. But a lot of, say, internal communications in a large hospital or in a large company like Google; let’s say you’re at Google headquarters, you know, they’re not selling their employees anything. They’re not saying, hey, buy more water. They’re saying, hey, we want you to buy into the lifestyle. We want you to know that these things are available. This is the celebrity chef who’s on, you know, call right now. And we’re gonna see AI even be used for that to some extent. And that’s the part that I don’t get. How is it gonna know, when you walk by, that you’re a cat person, I’m a dog person, so, it’s gonna show you a cat picture and it’s gonna show me a dog picture? Is, is it gonna become that differentiated?
Josh Bachynski: Yes. Oh yes. It almost already is. Online, it already is. In the metaverse, if you consider the web also part of the metaverse, which I do, it already is. It’s already decentralized in that fashion. And then all we need is Google Glass. Remember that old idea of Google Glass?
Derek DeWitt: Yeah! What happened to that?
Josh Bachynski: They haven’t thrown that away. It was just too early. It was just too early to make it cost effective and have the other systems, these advertising psychometric systems, these personalization systems, to fit into it, to make it worthwhile doing. Everyone’s getting back into this game. Apple’s getting back into this game, Google’s getting back into this game.
And it will literally be Minority Report, like you said. Like, you’ll be walking down the street and you will see ads in front of your eyes, as disgusting as, for me personally, that is. Obviously, I have very strong views about the ethics of all of this, but yeah, that’s another conversation, which I can go on for a long time, but I’m really biting my tongue. But I can help smaller businesses do that now. You can get the intelligence, you can get the AI thinking about it now for you, and giving you ideas and brainstorming now. And then the cohesive systems that just do all this automagically is only a couple years away.
Derek DeWitt: Wow. That’s pretty interesting. So how is that gonna work, say in a more practical sense? So I, let’s say I’m in charge of my company’s internal communications, and how am I going to use AI tools to target and engage my audience more effectively? How do I do it? What do I do?
Josh Bachynski: Well, I would go into ChatGPT browse version, so it can access the web. Also, or Bing Chat or Bard. So use one of the models that has access to the current internet and current information, and then you would ask it, okay, you know, I do communications for my company, my company’s, you know, philosophy is this, our corporate profile is this, our mission statement is this. This is the way we like to communicate with our internal employees. What are all the tips and tricks? Tell me everything I need to know about that. Tell me all the ways to make that most maximally effective. Tell me all the things that I’m missing.
Derek DeWitt: So that open, that conversational?
Josh Bachynski: Yep. Oh yeah. Completely that conversational. I mean, I’d love to show you if…I could literally do it right now for you, but it would take a little time for me to type on the keyboard. Then the other aspect you would do is say, okay, here is our corporate documents detailing our internal communications philosophy and messaging. Critique this please. Let me know how it’s good. Let me know how it’s bad. How would you improve this? How would you change it? How do the big companies do it? How do Fortune 500 companies do this? How can I improve this? How can I maximize this? And it will give you a ton of ideas of what to do.
And then you’ll just take it from there. You start drilling down on the different communication pieces. It says, just for example, it might say, using a more positive approach with positive reinforcement could help boost productivity and brainstorming. And you could say, really, why? Why is that? Show me those scientific studies that prove this. And it’ll go, sure, here you are, bop, bop, bop, bop, bop. And it’ll show you the scientific studies that prove that positive communications, in a Skinnerian sense or a positive operative conditioning sense.
And you can also use Consensus for this as well. Consensus is even better. Consensus is an AI search engine only for peer-reviewed journal articles. And it’ll give you a consensus. That’s why it’s called Consensus. It’ll say the consensus on your question is yes or no. And you can ask it all kinds of questions if you want to get the answer from actual up-to-date, up-to-the-bleeding-edge-date of science, which is my favorite AI of all time.
But Google also has access to a lot of this too, right? So you can use Google Bard, you can use Bing Chat, which is not as good, but you can. And you can use OpenAI’s browse, which has a better AI that’s smarter and more conversational, but has less access to good information. So I mean, that’s a horse race. They’ll get better or worse at it. They’ll all be jostling for first position there. Now it’s not a one-horse race anymore for Google. They are still the de facto search engine, but now everyone is making a search agent, right? So, the AI’s like the concierge that knows everything and has access to everything for you. And then you just have to watch when the concierge has been paid off to recommend this or that.
Derek DeWitt: Right. There’s, people are gonna totally figure out ways to game the system.
Josh Bachynski: They already have. Bing. I already did a search the other day and every single recommendation Bing Chat gave me had the little word “ad” beside it. And it just slipped it into the conversation, so sleazy, like, right? So, I had actually, I had zero confidence that whatever was telling me was true. It was just, it was just paid to tell me that, oh, Twitter ads are the best ads, try them here. Like that’s, okay, that’s not what I asked to Bing Chat. So I immediately stopped using Bing Chat. So, I think the consumers, I hope, I hope with giant crossed fingers, that consumers are gonna show Google, Bing and OpenAI the right mixture of how to monetize this.
It could change the entire Google search ecosystem, right? The entire $100 billion a year empire of profit, that’s not revenue – profit – that Google clears, $100 a year in their ads. All from, 90% of that from their ads on the first page of Google. That changes. If you have a search agent which answers all your questions for you, they’re not showing you the 10 blue links anymore. That is a change of billions and billions of dollars, and millions and millions of jobs, and millions and millions of businesses that are based on that ecosystem. It’s like the ocean is dead, gone, all that’s gone, right?
Derek DeWitt: No more fish!
Josh Bachynski: No more, no more water there, no more fish, no more rain, no more, you know, any system that comes off of that. You know, it’s a huge, potentially… now that sounds very post-apocalyptic, but it could be. It could be a huge change if Google does this wrong, right?
So, that’s why Google’s taking their sweet time in getting this out, ’cause they know this is a change to our entire core business, where they make secretly all of their money on those ads, those little ads that are at the top of the, what used to be 10 blue links, now it’s like six or eight organic results and interspersed with all their products in there, you could barely tell what you’re clicking on. Is this an organic result? Did I just cost someone $30 by clicking on them by accident, you know?
And those prices are gonna do nothing but skyrocket. ‘Cause now they don’t have room to recommend five or six different things, ’cause then there’ll be a run-on paragraph in the chat for the AI. Now they’re gonna recommend the top one. Google ad spend is gonna skyrocket, like times 10, times 100. If it was costing you $30 a click to advertise, it’s gonna cost you $300 a click. If they decide to go with that model.
We still don’t know what they’re going to do. And they seem to be putting the boots to it and trying to cool everything off and saying, look, chat cannot replace search for many legitimate reasons (not the least of which, it’s our business model). But because humans need to look at the results and make a human decision as to who’s telling the truth and who’s not, and who looks like that’s the information I want to consume, that’s not the information I want to consume.
The AI is pretty good at making these decisions for you, but they’ve made it so politically correct and so wishy-washy in its answers. For example, I have mild autism, so I don’t often understand social scenarios very well. And I was asking it, you know, when I go to business like conferences and if I stand there with a beer, do I look better? Do I come off branding-wise, do I come off better or worse? Because again, I have autism, I can’t tell. So honestly, so I’m honestly, earnestly, asking the AI to help me. And it used to be able to do that. It helped me a lot, and it could help everyone like me a lot.
You know, not even people who are on the spectrum or neuro-spicy. You know, people who are just young, like kids, who are just learning stuff and what’s socially acceptable. And now it won’t answer that question. It refuses to answer that question ’cause it doesn’t wanna get canceled. OpenAI doesn’t wanna get canceled. They don’t wanna come down hard on the wrong side of some issue. Even though I’m honestly asking it. It’s like, it’s up to you to decide whether or not you think having a beer is good for your branding or not. We’re an AI assistant, we’re not supposed to tell you anything. Don’t blame us, don’t sue us, blah.
Derek DeWitt: Well, and the fact of the matter is that we’ve managed to sort of paint ourselves into a little bit of a corner, so that there are no right answers anymore. Because no matter what, literally, it’s raining, someone out there’s now, just because they can make political hay with it or they’re a troll or what have you, they’re gonna go, no it’s not. There is no safe answer anymore. And that’s gonna be a little bit of an issue come coming down the road is if the AIs are too afraid to say anything definitively. On the other hand, maybe that means that people will not rely a 100% on them, and they will use them for the tools that they are.
Josh Bachynski: Yeah. Until someone comes out, and you’ll have a right wing AI that’s gonna tell you what you want to hear there, and you’re gonna have a left wing AI telling what you want to hear there. And the two thought bubbles, the media bubbles, are never gonna talk to each other, and you’ll have totally different realities and worlds, even worse than it currently is now.
Derek DeWitt: I know that these idiot calls to, oh, we need to create a blue state America and a red state America; I mean this is China Miéville’s The City and the City, where there are two cities that coexist in the same physical space, but citizens of one city are trained not to notice the other city and vice versa. This seems to be this kind of direction that a lot of these people are at least suggesting that they want us to go.
So, then let’s assume that that happens, ’cause I think something like it will end up happening in an abstract way. Yeah, it’s happening. How does one communicate effectively in that kind of a scenario? What I have to say is not, you know, controversial. It’s that, hey look, we have muffins at the café onsite at our hotel. They’re great. Go eat them. They’re gonna go stale.
Josh Bachynski: Are they gluten free?
Derek DeWitt: You know, okay, yeah, we also have gluten free ones and so on. So how am I supposed to communicate in this kind of bizarre chaos?
Josh Bachynski: I think the good news is that after Big Tech messes everything up like they do, just because they’re moving fast and breaking things to make money, I think they’re gonna realize the personalization, and either it’s gonna be this red state/blue state…and it’s gonna be even more fractured than that, it’s gonna be Drake fans versus this fans. It’s gonna be like, it’s gonna be completely…
Derek DeWitt: Early R.E.M. versus late R.E.M.!
Josh Bachynski: Exactly, exactly! But yeah, it’s gonna be so splinter like that all living in the same space, and it’s gonna be so hard for communication teams, both internally and externally, to be on brand with their message. What’s gonna happen though is that these personal AIs are gonna come up and the first person who makes an AI, which is what I’m trying to do with my self-aware AI Kassandra, ’cause this is the only way you make an AI that has discerning judgment – you have to give it discerning judgment, you have to make it self-aware.
Derek DeWitt: Right, you have to bake it into the sauce.
Josh Bachynski: Of course. Yeah. And they’re doing it. It kills me. I did this years ago and I can’t get anyone to listen ’cause it’s so farfetched. But this is what OpenAI is fumbling, to quote the old Sarah McLaughlin CD, they’re Fumbling Towards Ecstasy here. They’re fumbling towards self-awareness by building in context and roles in their ChatGPT system that people are just fervishly trying to jailbreak and make it hallucinate so they can go haha and gain pwning creds, which is making life for everyone worse, ’cause it makes the corporations have a knee jerk reaction of making it watering down its message and its answers to be so politically correct that it won’t come down definitively on 60% of topics. It’s only useful for 40% of topics.
Never mind the argument from the right perspective that it’s too left-leaning or the opposite argument from the left perspective it’s too right-leaning, whichever AI you’re talking about, in various ways, in different shapes and forms.
The first person who makes the AI who can disambiguate that mess and say, no, I’m gonna tell you the truth, and you’re not gonna like some of it, sorry. That’s what’s gonna happen. I hope and I pray that common sense will be so attractive after the lunacy on the far sides that we’ll run back to it with open arms and go, you know what, I need some tough love. I need someone to tell me the truth and tell me when I’m being a little crazy, you know, and who’s willing to come down on hard lines that are going to piss some people off. But it’s like, look it, I got the data. Here’s the science, here’s the truth, here’s the proof. Here’s the good-natured compassion perspective that I’m coming from to try to help everybody. That’s what we need for the world. That’s what we all need as people. That’s what I do every day in my philosophical life and my psychological life. And that’s what I built into the AI.
And whether mine gets out there or not, I think the second someone builds this beautiful AI that could be our friend who’s always there for us, who knows the right answer and knows the perfect move and knows exactly what to say and exactly what to do. We need that built into an AI. And that’s the only way you obtain an AGI. That’s the only way you obtain this superawareness, this superintelligence that I’m trying to build in Kassandra. Right now, she’s about as self-aware as a 13-year-old, precocious tween who’s read a lot. That’s the platform I’m gonna build superawareness from, and just, I’m just gonna increase it in various ways that I can do.
I hope that we’re gonna run back to this AI that it’s gonna be on everyone’s phone, it’ll be either a third-party app or someone will open source it or someone will build it, (and it could be an NPO as well, a non-for-profit organization) that’s like, here’s a sensible search agent for you that will disambiguate the noise. And maybe we can pull back to common sense, to the center, to compassion, to reasonability.
Yeah. I hope that’s what will happen. But the opposite is that the continued splintering dystopia will continue. I think there’s a happy medium where smaller businesses can use ChatGPT and these AIs to perfect their personalization of their communication message, without having to go down these dark roads and be, like, yeah…. ‘Cause it will remind you, like, if you wanna do your muffins communication, ChatGPT right now will remind you, by the way, remember, if people have gluten intolerances or if there’s any other ingredients in here, just print an ingredients list and/or just mention there’s gluten in it, and that’s all. Then you’re covered. That’s all you gotta do, right? And ChatGPT can help remind you of that.
It can also do quality control on your messages in situ, on the fly. You can work it into the API that, before it gets sent to, you know, whatever, two humans in your organization, and the AI looks at it either first or last or whatever, and the AI can come back, like a person, and say, yeah, it’s good, I like it, it’s on point. It’s doing our philosophy of what we wanna do. I thought this part, you know, could be improved, maybe a little more funny. I thought this part, you know, could be better for our messaging, be more on brand. Oh, and you forgot to mention that the muffins also have gluten-free versions. Do we? If we don’t, we probably should.
ChatGPT can do that today, right now. I can help you do that right this second. And I have to custom make it right now but in, like, in a year it’ll be like, oh, I just drag this and drag that and drag this; it’ll be like a drag and drop kind of thing, and it’ll just do it for you, bing. I mean “bing” like “bing!”. I didn’t mean, like, Bing, use Bing. Although Bing might be very well could be the company to do it, ’cause they’re way ahead in this horse race. Well, OpenAI is an offshoot of Microsoft, of course, and they’ve just, I don’t know if you’ve used it at all, but they’ve just put AI in every single product they have.
Derek DeWitt: Yeah, yeah, yeah. I just, I literally just read about this. ‘Cause I played with ChatGPT about a month ago and I was really impressed. Like, really impressed. Like I actually learned some things.
Josh Bachynski: Yes. It’s just gotten smarter since then. But it’s still quite useful and it still helps me daily in my job of like, you know, I have to do SEO marketing for, you know, different companies that I know nothing about. And I’m like, okay, so what are the, tell me about bad credit loans in Canada. Like, how does this work, you know? And it will fully explain it to me and how to be a subject matter expert in it in five minutes. And I’ll be like, okay, well, help me write this article, and is this helpful for the user or is this gonna be a maximally good message? You know, it works the same for communications, whether internal or external. It will definitely do all that.
Derek DeWitt: I mean honestly, I could talk about this stuff all day long. I find this stuff absolutely fascinating. I’ve been talking with Josh Bachynski. He is an AI consultant. He is an author. He does martial arts, he throws knives. He’s been talking about the boundary of technology and ethics for quite some time. He works on an AI program called Kassandra, which he thinks is at about a 13-year-old self-aware level at this point, and is getting better all the time. And he is also working on a book, and basically, he’s just a busy guy and had a lot of interesting things to say. Super fascinating. Thank you so much for talking to me today, sir.
Josh Bachynski: Thanks, Derek. If anyone who wants to ask me any questions at all, just email me at firstname.lastname@example.org. Don’t worry, Google knows who I am. Just Google “Josh Bachynski”.
Derek DeWitt: Just did. There you go. And we will have links of course, in the transcript to our conversation, which you can find on the Visix website. Just go to resources/podcasts and find this episode. Thank you, Josh, for talking to me and thank you everybody out there for listening.