Artificial Intelligence Meets SEO: Danny Richman's Unbelievable Journey

Mark A Preston: Welcome to the unscripted SEO Interview podcast. Yes, it's 100% unscripted, 100% unrehearsed, 100% unedited, and 100% real. I'm your host, Mark A Preston and today we have joining us Danny Richman, a consultant, brainer and mentor, and he describes himself as being a little bit obsessed with solving problems, using AI. So I'd like to welcome Danny. Hi, Danny.

​

Danny Richman: Hi Mark. How are you doing?

​

Mark A Preston: Very well, thank you. For the people listening to this and watching who don't know who you are, could you give a bit of an overview of your background and the history and your story of, where you've come from and where you are now?

​

Danny Richman: Yeah, sure, yeah. I won't spend too long on it, but basically I've been involved in tech now for about 43 years, and I started a software company back in 1983. So probably before most people listening to this were born and I grew that company and sold it in the mid-90s and thinking at that time, that I didn't really want to be involved in business anymore, in tech anymore, I was going to go and often pursue more creative endeavors, like music and acting and things like that. But then the web came along and Google came along, I just got sucked right back into it. I just found it fascinating and that's when I kind of started developing an interest in SEO, and SEO seemed to me like the perfect marriage for me because I was always interested in marketing and people and psychology, and then there was this blend between that and technology, and so it just seemed like a really good fit for me. And so initially, I started doing SEO consultancy, providing SEO as service to clients, but then realized that I actually preferred to do more training, so working with in-house marketing teams and helping them implement SEO themselves. I do still have a very small number of SEO clients that I work with, but I don't take on too many of those projects because as I'm sure you know, Mark, it's time consuming.

 

And so, that was me kind of ticking along and I had every intention of retiring about a year ago, and I actually managed to kind of partially achieve that. I started reducing the number of days I was working, I've got various other hobbies and volunteering commitments, so I was going to start moving in that direction and then AI came along and I got sucked into that. And, yeah, it very much got to the point of an unhealthy obsession in that AI became really all I could think about. It was whirring around my head, every given hour of the day. When I went to bed at night, I was thinking about AI; When I woke up in the morning, I was thinking about AI and I think I'm past that stage now, thankfully, because it really wasn't good and I've now kind of settled into a nice way of things. But for about the last year or so, I've just been building that's really all I've been doing is like, working with different businesses and organizations, building stuff, helping them to find practical applications of SEO, how it can help their organization in some areas, even create completely new products and services using AI, we might touch on that later. And yeah, working with them to get those things built and building stuff myself. So I am now working harder than I ever have in my life.

​

Mark A Preston: So that's what semi-retirement looks like, is it?

​

Danny Richman: Yeah, but Mark, I'm knackered. I'm absolutely knackered. Yeah, I mean, I've got a little child as well, who gets up at 05:00 in the morning, so I'm up at five, I'm working all day, I am exhausted, and I desperately need a holiday. Anyway, that's the path I've taken, that's where I am. But, look, I'm complaining, but I absolutely love it. Everything I do and have done, certainly for the last 30 odd years, I would do even if nobody paid me to do it. I don't tell my clients that. But, yeah, it's my passion as well as my work.

​

Mark A Preston: Yeah, I understand it. I always say that I'd do this for no money, if I didn't have to feed the kids.

​

Danny Richman: Yeah.

​

Mark A Preston: When it comes to SEO and AI and chat GPT, I mean, I think the SEO industry has got so fixated on AI being the content solution, rather than the many problems it can solve. I mean, what's your thoughts relating to what the SEO industry mindsets in at the moment?

​

Danny Richman: Yeah, I have heard from various people, who are now using GPT to create content and do it on the cheap, basically, so they now don't have to pay for content writers to produce that content. I have to say, I'm not a big fan of that approach and I think even if you are capable of getting decent results out of GPT, and I'm careful to avoid using the term chat GPT, because I think there's often a conflation there where people now think because chat GPT became so popular, people think that kind of chat GPT is AI or chat GPT is GPT or chat GPT is a chat bot, it's just a user interface. The underlying technology under chat GPT is what's interesting to me. And I think even if you can use that to create content, well, the thing is, so can everyone else. And so how far is that really going to get you, if all you're doing is churning stuff out with the machine that every Tom, Dick and Harry is going to be capable of doing as well at almost zero cost? So, I don't know that it's really sustainable as a business model to be running all your content marketing based on AI generated content.

 

So, it's not something I would recommend to people. Absolutely use it as a tool, like a writing tool, much as you would something like Grammarly to enhance your writing. You can use it for structuring your content, coming up with ideas for content, there's all kinds of really good ways of using it, so it's a helpful tool, but to completely replace the writers in that process, I don't think that's a great idea; I really do not.

​

Mark A Preston: I was at Brighton SEO, last year watching your talk relating to something you'd come up with a proper use of using this technology, within helping to streamline everyday life within the industry. Is that how you see yourself focusing on creating something, shapes or tools to help people solve problems?

​

Danny Richman: Well, when GPT first came out, I started developing an interest in it. I was actually looking the other day, when did I first start playing around with this? And my first tweet about GPT was in June 2020, and I hadn't realized it had been that long. It seems like no more than a year ago, but yeah, it's incredible; like, three years has passed now. But when it came along, obviously I was experimenting; I wanted to see what it could do, and I wanted to try out different ideas, and I wanted to use it to manipulate and play with data and see what it could do with some data that I had. And obviously, because I'm working in SEO, most of the data I had was keyword data or audit data coming out of tools and that kind of thing. So I wanted to see what I could do with it and that data was sitting quite often in a spreadsheet. And so in Google Sheets, you've got a programming language in Google Apps Script, which sits behind the sheet, you can make an API request to OpenAI and call these models in, manipulate the data, and do all kinds of cool stuff with them.

 

So originally I released just, I mean, everything I did, as I was finding practical uses for it, I just thought, okay, well, other people are probably going to find that helpful too and so I just released them out. So some of the things I did initially were, for example, let's say you're doing some keyword research. So the example I gave at Brighton SEO was, I was doing some keyword research for a law firm, and we were looking to create some content around various different conspiracy charges. So you got things like conspiracy to murder, conspiracy to supply drugs, all these kind of different things. Now, if you go to a tool like Ahrefs and you put in the term conspiracy because you want to see everything, right, 90% of the keywords it will spit out have nothing to do with the law. I mean, there's like UFO conspiracy, Diana death conspiracy, all this kind of stuff. So, how would you then go through that list of possibly 10,000 or more keywords and pick out only those that have a relevance to the field of criminal law? Well, prior to GPT, it was some poor sucker's job to have to sit and go through that data, ticking boxes or trying to add filters into Google Sheets and all this kind of thing. I just ran the whole thing through the GPT API; job done. It was just perfect and it did it really quickly, so I think that was about one of the first ones I did. Then I started playing around with some data. Do you know that tool SparkToro, from Rand Fishkin?

 

So with that tool, you can do some audience research and you can go and have a look at all the topics that your audience, your target audience are talking about, tweeting about, their hashtags, all this kind of stuff. So I then took that data, pulled it into a Google Sheet, and then got GPT-3 to create content topics based on all the things that they're talking about. So that was another one I did. See it's a while ago now, so I'm not even sure I can remember all the things. Oh, another one I remember was, do you know the platform? Hey, Row, help a reporter out?

​

Mark A Preston: Yes.

​

Danny Richman: Right. So, they send you all these kind of journal requests, where the journalist or the blogger is looking for a source for a story, and they outline all their requirements of the expert they're looking for to contribute to this story. Well, who's got the time to go through all of those requests, figuring out whether you are the right kind of person for that story, and then having to write pitches to all these journalists to try and to be included. So again, just use GPT to then pass through those emails, extract all the information, work out whether it was a suitable match for you or your client as a source, and then write to the journalist outlining exactly why you are the best source for that story. So it was just one after another, but where I think there is a really useful application for AI in SEO, it's not one that I will be doing because it just doesn't interest me that much to do it. But you know how if you do an audit of a website and you use a tool like Screaming Frog or you might use SiteBulb or something like that, and those tools will then come up with a whole list of potential issues with the website, hundreds and hundreds of different things. Some may be important, some may be not important, all that kind of thing and I think there is a real opportunity for someone to build an integration into a tool like that and maybe they're working on it, where it looks at those issues that are thrown up and just fixes them on the fly.

 

So you could actually have those pages modified on the edge, so that you don't even need to make changes to the underlying website content. They could be served to the client with those changes. So if, for example, your Ahref lang tags are all messed up, you haven't got the right links to the right proper pages, it would be a fairly simple task for AI to look for those kind of issues and then make those changes on the fly and serve them to the client, as the site is being crawled or indexed.

​

Mark A Preston: When you say serving on the fly, do you mean giving them the information of what to change or actually getting AI to change it on the website itself?

​

Danny Richman: Yeah, getting AI to change it on the website, but it doesn't even need to make those changes on the actual server, where the website is being hosted. I don't know if you're familiar with this concept of edge computing, where when a client makes a request to receive a page, to get the URL of a page, that content, that page, the HTML and all the code that's behind that can be delivered from an edge server, and that could just be adapted on the fly as each page is requested if there are any issues in that page and looking at it within the context of the whole website. I mean, that would be quite a big project; there would be quite a lot, technically to look at there, but it would be a very interesting one. And I'm sure at some point, either someone's working on that right now because I've been talking about it for a very long time, them as have a couple of other people. I'd be amazed if someone's not sitting somewhere working on that problem at the moment. But I have to say, pretty much all the things that are taking my focus at the moment and the projects that I'm working on don't really have anything to do with SEO. I've worked on recently, three big AI projects, and none of them have anything to do with SEO, they're all in other areas. My attention has kind of been taken away from looking at this technology to solve problems in the world of SEO and looking at it in other fields.

​

Mark A Preston: So, what kind of problems could you give me a scenario and what kind of problems that you're using AI to help solve?

​

Danny Richman: Yeah, so there's some of them I can talk about, some of them I can't talk about in a huge amount of detail, but I can talk kind of in a broader sense. But one of them I can talk about in detail and in fact, I even wrote a case study of it up on my blog. So it was for this law firm, same law firm that I was working with on SEO. So one of the challenges they have in that firm is, they are very busy, very successful company, and they receive somewhere around about two to three-hundred new client inquiries every day. And most of those are coming off their website and most of those are people filling out the inquiry form on their website, and these are all people who have they're in trouble, they've been arrested or charged with some kind of criminal offense and when they fill out that form, quite often they're putting their whole life story into that form. They're saying everything that's happened up to the point of them now sitting in a police station somewhere, right? So it's the whole story. Believe me, you would be shocked at some of the things that people write in here.

 

Anyway, so this law firm, they had a team of three people whose sole job was to look at all these inquiries coming in every day, read through them, work out whether the person making the inquiry was accused of a crime or was a victim of the crime because some people don't actually realize what criminal defense is. So that was the first thing. Then, assuming they are being accused of a crime, they then had to work out ,what crime are they going to be charged with, what offense have they committed? And then once they've worked that out, they then have to work out whether the offense is serious enough to go to a Crown Court trial, and the reason they're doing all of that is because if it is a serious offense, that means very big potential, potentially very big fees for the law firm, and they want to make sure that they respond to that inquiry within five minutes of it having being made. It can't wait, because the quicker they respond to that client, the more chance they've got of securing that client. So that was their job. This team of three people were just doing that all day, making these evaluations of the inquiries and then once they've done the evaluation, they then route it onto whichever lawyer is most applicable to dealing with that kind of case. And I was talking to the head of this company and the law firm, and I said to him, I think I could maybe get AI to do that task. And so I set up a system for them that did exactly the same job as that team evaluates the inquiries, classifies them, routes them through to the right lawyer. We ran it as a trial for three months, comparing the evaluations of the Human team against the evaluations of the AI team. And we found the AI to be significantly more, more accurate than the Human team.

 

So all of that team were then moved into roles that allowed them to actually work on client cases, nobody lost their job. They went into working on client cases, a much more interesting job than having to sit and shift through all these inquiries. So that was one, it's up and running; it's working really well for them; it can deal with those inquiries at 03:00 in the morning; it never needs a break, it doesn't need a holiday, it doesn't get sick, and it's very scalable as well. So that's kind of one example. Another one, which I'm going to have to speak more broadly about, was a large American university that have an investment fund, and they receive a lot of pitches from fund managers and startups looking for investment. And so they then have to again, it's very similar to very similar to the law firm thing. They're basically taking these pitches in and trying to evaluate whether this is a pitch that they would be interested in investing. And that, again, required a whole Human team to evaluate all these pitches, categorize them, make decisions on them and we just got AI to do that instead. And then the one I'm currently working on at the moment, which should be publicly launched in about three to four weeks, is, how can I talk about this without giving too much away, but it's essentially, essentially using AI to allow customers to completely create their own product from scratch. Oh, that's as close as I can get to description.

​

Mark A Preston: So, theoretically, you are getting a machine to do the job of what a team of humans are doing now.

​

Danny Richman: Yeah or sometimes getting a machine to do a job that the human could never even do. It would never have even been conceivable. I mean, like this one I'm working on at the moment, it just would not be viable to create this product uniquely for every consumer according to their specific requirements and to make that cost effective and feasible without this technology, it just couldn't have been done.

​

Mark A Preston: So, what's your thoughts on all these people out there that are saying, well, AI is responsible for taking people's jobs?

​

Danny Richman: It hasn't been my experience that's what's happened. What I have found where I've been involved in projects, is that nobody's lost a job. People have just been moved into different types of jobs and certainly historically, when you look at the impact of technology on employment generally over the years, this isn't the first time there's been a big technical revolution. Historically, it just hasn't been the case that you've got millions of people made unemployed, you've got certain people doing certain jobs that are no longer needed. There have been whole sectors, whole industries that have been displaced as a result of technology. But certainly in the UK, we do not have huge levels of unemployment. So, you know what's happened to all those people? They've just gone on to do different things. So again, it's hard for me to make predictions and I don't know what the future is going to hold, but my hunch is, that's what will happen this time as well, is that we won't necessarily see millions of people sitting around with nothing to do. They'll just be doing different things, they'll be doing the things that only humans can do.

​

Mark A Preston: So moving things back onto SEO for a moment, speaking about the industry as a whole, how can the industry use AI to help streamline, what they're doing in a more positive way?

​

Danny Richman: Well, I'm kind of reluctant to answer that question because, I have a very strong feeling that the whole world of SEO is going to change enormously.

​

Mark A Preston: Yes, and I definitely agree with you there.

​

Danny Richman: It's only yesterday, that you had this Google IO conference where they were talking about what they're doing, and I have to say I'm active on Twitter, as are you, and most people in SEO kind of busy on Twitter and I look through my Twitter feed and I see all these people kind of putting out messages and posts and things that are not really any different than they were two or three years ago. It's all on the same sort of topics. And I just find myself thinking, your whole world is about to change, the whole field of SEO, the whole field of search is about to change. All of this stuff that we've been doing for the last 20 years, just seems to me a little bit redundant now. And I think that for me, I prefer to train my thoughts less on his stuff that we've always done; how can I use AI to solve a problem that we've always done and more about using AI to do things that we have never done before, and that for me, is a lot more interesting. And I think that's true of AI generally, a lot of people are just saying, oh, I used to have to write stuff; now I can get AI to do it. I used to have to do this process manually, now I can get AI to do it. For me, it's much more interesting to look at ways we can use AI to do things, we've never done before in completely new and different ways. Look, I'm sure if I sat down and that's where my focus was, I'm sure I could think about all kinds of different tasks, that people who work in SEO are doing every day and find ways to streamline their work with AI. I have to say, it just doesn't interest me and I think it would only be something that would be useful short-term to people, because I think search is about to change beyond recognition.

 

And I know a lot of people disagree with me, a lot of people say, oh yeah, AI in search, it's just going to be like featured snippets. It's just going to have a bit more impact on the SERPs. I really don't think that's true, I really don't think so and if you have a look at what Google put out yesterday and how that search page is changing, I don't even think search is search anymore. I think search is a concept that will no longer exist this idea, that a search engine is a way for people to go and find information on the web and then show a set of ten blue links to get the results. That's gone, that really is. Go to Google today. You might still see something looking like that or close to that, but I don't think that's going to be around for very long, because the fact is that Google users, search users, I don't know that they care much where the information is coming from or how it's got to them. They just need answers to their questions. They want to know the best place to go and buy something. They want to know whether this bicycle has the right features that they're looking for, they want to get answers to medical questions that they've got, all kinds of things that people use search engines for. Now, if the search engine can just answer that question for them, doesn't need to show them links, it's just going to answer that question for them. Not only answer it, but then I can go and enter into a dialogue with those results to then get more information, have things explained to me in a different way, expand on it, simplify it, you know, test me on it. Why do I need ten blue links? Why do I need them? Only SEOs need ten blue links. All these conversations about the way we've always done SEO, just for me, just seem a little bit pointless now.

​

Mark A Preston: Yeah. Reminds me of the SEO industry suddenly, all at once over the past year, suddenly AI is this new thing. I mean, I did a keynote talk, what, five and a half years ago now, on AI, and my exact words was, if you're only just thinking about AI, you're already way behind time and that were over five years ago. But I think the industry they just focus on here and now, whatever's happening today, that's the focus and they don't try and seem to predict or look into the future of what's going to happen.

​

Danny Richman: I think a lot of the way that AI has been used up until now, certainly by Google, has been much more on the back end, where they're refining search queries and they're using it to understand searcher intent and the context of a search and all those kind of things. So I think that maybe SEOs have felt less threatened by it because of that. I'm guessing they probably almost certainly used it, and they must have used it as well for doing things like checking spammy content, checking link manipulation, all those kind of things they must have been using in those areas, but it was still on the back end. And I think that the big change now is that a lot of this is going to be on the front end, and the notion of ranking number one on Google is going to have a lot less meaning or even being page one on Google will have a lot less meaning and what I found and what I've experienced is that there seem to be kind of three different categories of people, that talk about AI on social media. There's the people who are just like, wow, this is great; it's so exciting. Just think of all the things we can do with it. Bring it on, there's that group. Then there's those that say, this is going to kill us, or it's going to wipe the whole human race.

 

This is the worst thing that's ever happened, that's the second group. And then there is a third group, which is I call them the dismisses, which is, yeah, yeah…yeah, AI it's not that great. What's no different to featured snippets? It gives inaccurate answers. It's not that great. I misunderstood that group for quite a long time because I thought that they were really only looking at kind of what they could see in front of them, rather than thinking a little bit a step ahead. But what I've come to realize is that the dismissers are actually the same as the people that are fearful; they're both afraid, people are afraid. There's a lot of people that are really worried about the impact AI is going to have on their agencies and their livelihoods and all kinds of things. So I get it; I really do. And I don't like the idea that people feel worried and I don't know how much reassurance, I can give those people but all I would say to those people who are concerned is that, if you didn't want to be in an industry that was constantly changing and evolving, then you probably should never have been in SEO in the first place because this is a very dynamic industry and it's just like constantly things change and move and shift and that's the nature of the beast. So this is another thing that we'll have to adapt to and understand and see where we fit in. What part do we play in that?

​

Mark A Preston: Yeah, I think the change that the industry is going through at the moment and what's been going on in Google and everything, I feel personally that it's one of the most significant changes that's happened, especially in the 20 odd years I've been in the industry. I feel as though this is like that, “wow”. The ultimate change of all change is nothing. People are not dismissive. Yes, they're very much afraid, but they're like, what do we just do? What is the answer? I think that's it —working out what the answer is.

​

Danny Richman: That's a natural reaction because when people are worried, what they want to do is have some sense of a plan in their heads as to, well, how do I adapt? how do I survive? This is a very natural and primal human instinct for your head to start getting into that space as to what do I do? and uncertainty is very uncomfortable for people. Not knowing what's going to happen in a couple of month’s time, not knowing what your work is going to look like and what your industry is going to look like is very unsettling for people. So I really do understand it but there's a saying that's been kind of doing the rounds in the AI sector for a little while now, where people have been saying, AI won't take your job; someone using AI will take your job and I think there's a lot of truth in that. And I think probably the best thing that anyone can do and this doesn't apply to only people in SEO, I think the best thing that anyone can do right now is to don't dismiss this technology; try not to be fearful of it, but learn to understand how it works, what you can do with it, how you can use it to your advantage. I think it's the absolutely best thing anyone can do at the moment in almost any field. And I think when I think about some of the people whose jobs might be at risk, I have to say, people working in SEO are not first and foremost, in my mind. I think this technology is going to be hugely disruptive to so many people in so many fields of work. I mean, these lawyers that I'm working with, I can very much see a time in the not too distant future, where it's going to be questionable why we even need those lawyers.

​

Mark A Preston: WOW. Just thinking about the opportunities that's out there and I think it is about refocusing the energy, working with something rather than against it and looking at it, yes, the customer service is already being massively hit. There's people's jobs being lost because of AI, but ultimately looking at it on how we can use it as part of our team rather than the competition.

​

Danny Richman: Did you see the thing that was going around Twitter today about Wendy's The hamburger restaurant are now using chat bots, but spoken voice chat bots as customer service agents? They're taking people's orders, handling inquiries, all this kind of thing. So you go to a drive in at Wendy's Burgers in America and Wendy's have said they've now been trialing it and they've said that, it is a significantly better experience for their customers than the humans that used to do that job. I think there will be many different areas that are going to be disrupted here, but I think for as many jobs that are lost and people that are replaced, there will be at least as many new opportunities created, if not more. I am doing things ,I said to you, I've been working on a lot of projects and I've been building like crazy, right? Well, yes, I used to be in the software industry and yes, I used to be a developer, but that was like 30 years ago since I did any coding. My coding skills are so out of date. I mean, most of the modern languages, I have no familiarity with at all. I haven't done that stuff for a long time, I am a terrible coder compared to anyone doing that. Well, I've just built three fairly significant applications and I've just got GPT to write all the code for me.

​

Mark A Preston: Now, where does that fit with the developer? So, traditionally you have developers who are there to write all this code. Do agencies and tech teams, do they still need those developers because surely you need somebody that understands, what AI is writing code to understand if what it's doing is the right thing.

​

Danny Richman: And I think that had I not had a background in software development and I didn't have some understanding of the principles of code, I don't think I could have used these tools as successfully as they have been. I need to be able to understand the logic of the code. I need to understand when things don't look right, where I think it's misunderstood my requirements. When errors come up, I need to understand what might be causing that error. Although, having said that, what tends to happen is, I give my requirements to GPT 4, it spits out the code, I run the code, it comes up with an error, I then put the error back into GPT 4, it then tells me how to fix the error and that's basically the process that we just keep going through. So it's actually much better at debugging software than I am. But I can tell you that the reality is that all three of these projects that I've just worked on recently, since the beginning of this year, had we had to go to traditional developers to go and build these things, these projects would never have happened. I mean, the cost of that development would have run into several tens of thousands of pounds, and they would never have happened. They just wouldn't have happened. So no developer has lost a job as a result of those three applications that I've been involved in. They're just new opportunities that have been created as a result of being able to do these kind of projects without need for a developer anymore. And that's exciting, because now there is pretty much no friction, whatsoever between you having an idea and then realizing that idea and getting it out into the world. Now, that, for me, is thrilling. Just this whole idea, that I can lie in bed at night, something sparks in the back of my brain somewhere, and I think, oh, there's something interesting, and I can make that real the next day at zero cost. How exciting is that?

​

Mark A Preston: So, would you say that AI has become something that people can create ideas into reality at a much reduced budget than traditionally they would have to?

​

Danny Richman: Yeah, I mean, a really physical and visual example of that is, if you have a look at tools like Midjourney and these image creators my artistic skills and photography skills are really poor, but I have been producing imagery for all kinds of different applications and different things, beautiful imagery that you would never, ever know was created by AI. And I have no artistic skills whatsoever, but I have the ability to have an idea and express that idea in English and then get AI to create it to me, create it for me. And so I think that's just a really clear example for people to see how, my imagination; you can just really let your imagination run riot. I was actually working I did some training yesterday with someone from a company, and I was working on helping them develop a content strategy and I could see that this woman I was working with was quite anxious about it. She was quite anxious about out the idea of coming up with content topics, understanding what to write about, how to structure that content, what form it should take. And it was supposed to be a session about SEO and then I said to her, do you know anything about AI?

 

And she said, I've heard of it, I've heard about artificial intelligence. So I said, have you seen something like about all this stuff that's coming up with chat GPT and all this kind of stuff? No, never heard of it. So I said, what about things like Midjourney and stable diffusion? No, don't know what you're talking about? And so I showed her and I started went to GPT we took one of the content topics that we put in and got it to develop a structure for the content and then write some paragraphs and get prompting it to kind of get, I mean, her mind was literally blown away. She thought I was doing some kind of sorcery and it was such an incredible feel, healing, to introduce that to someone for the first time, to show them what it could do and create these images and do all these things and I could see that as soon as I had kind of given her this gift of knowledge, to see that this is what you can do with this, it was just like, “Wow, Oh, wow.” And she just got so excited with all the possibilities of what she can do with that. And then a few weeks ago, I went down to my local junior school, and I gave a talk to year six, who were like 19 years old, and most of them had never heard of AI; they had a very, very vague idea. Some of them had a vague idea of what it was. And I think the teacher had also prepped them in advance of me coming down, and we were doing things like I asked the teacher, she was a great sport. I asked her to give me a sample of her speaking for like, a few seconds.

 

So I created an AI model of her voice, and then I took a picture of her and created this video model of her in her voice and was getting the kids we were typing in. I said, tell me what you want your teacher to say and so we were getting I mean, some of it I had to kind of filter because it wasn't appropriate but they were getting her to say all kinds of things like, hey children, the whole class can have the rest of the week off. You can all play on your Nintendo switch, all this kind of thing. And they were so excited and thrilled. We were doing things like we were taking pictures of the playground, and then we were doing in painting, where you kind of mask off an area, and then you say what you want to have in that area. So we were putting, like, dinosaurs in the playground, Mario brothers in the playground. And then we were doing stuff like so they're working on the current project they were working on was the theory of evolution, Darwin’s theory of evolution. So I created a Darwin bot, and so they're now having a conversation with Charles Darwin, asking him about his work, his experience, his life story. I mean, wow, if only I had that when I was a kid at school rather than those dull lessons that we had to sit through. So, yeah, I can only feel nothing but hugely excited for the future and just completely engaged with this technology and all the possibilities of the things that it can do now and will be able to do for us.

​

Mark A Preston: So when it comes to this technology and whether it's imagery, video content, code, whatever it is, who actually owns the copyright of what you create?

​

Danny Richman: Nobody.

​

Mark A Preston: So basically say you create a tool. Do you own that tool?

​

Danny Richman: Well, are you talking about a tool like a software tool? Like I've used it to create?

​

Mark A Preston: I'm just trying to get the head round of not so much legalities, but who actually owns it because from my understanding, you own what you put into it, but what it spits out at the other end actually owns that.

​

Danny Richman: I'm not a lawyer here, but my layman's understanding of copyright is that copyright relates to a human creative endeavour, something that you as a human have created. Now, I think it's simpler to get your head around it, if we talk about things like text that has been generated, like an article right, or a picture that you created with Midjourney, rather than talk about software, because we start to go into different areas there because the software gets a bit more complicated, because there's an argument that even traditionally written software doesn't have copyright there. It's not seen as a creative work. So let's take an article that you've generated using Chat GPT, right? No human has created that article. There is no creative human endeavor involved in that article. This has never been tested in the law as far as I know yet, but from what I have read and researched, nobody can claim to have copyright in that work. Now, if you take that article and adapt it and you start chopping and changing it around and you're going to start putting in your own commentary, in and amongst it, well, then you're creating something new and in that you may have copyright but if you're literally just taking stuff straight out of Chat GPT, nobody owns the copyright and the same with images you create with Midjourney nobody owns the copyright.

​

Mark A Preston: Right. What about all these people that say, well, these AI tools are actually just basically basing their information on what other people have already written or done?

​

Danny Richman: Yeah, so I think there again, quite often the people that make that comment, don't really fully understand how these tools are working. All like GPT does, is it takes this huge corpus of data, this training model that's been fed with, which is millions of web pages, Wikipedia, all this kind of stuff that it's found, and it's not using that content. If you ask it a question like, write me an essay on my trip to France, it's not going and looking through that training model, trying to find an essay on a trip to France, and then pulling that and then putting and then giving you that as a response, right? It's using the training data to understand what all these concepts are? What does France mean? What does Paris mean? What does a trip mean? What does a baguette represent? All the kind of things. So it's looking at all the things that have been written about people traveling to France and understanding all these concepts and then creating something new and original from that model. And what it's creating is not even necessarily designed to be accurate; that's not its job. It's designed to be probabilistic; that's its job. It's designed to give you the most probable answer. In fact, I did a little demo of this, where I don't know if you're familiar with a concept called embeddings, where you can basically upload some data into GPT and supplement its training model with your own data.

 

And so I did a little example of this and posted it out on Twitter a few months back, where I had a Google sheet and it had a list of 25 different types of animals on it. Lion, bear, snake, all this kind. And then in a cell in the Google sheet, you could type in a phrase. So, the first example I gave was howls at the moon and I had created it so that it could only choose from the list of animals that was in the Google sheet, could only give an answer from that, had to give an answer, but only from that list. So I put howls at the moon, it came back with wolves. So what's happening here? It looks at all its training data, looks at which of these animals on the list is most commonly associated with howling at the moon, and comes back with an answer of wolf. Okay, makes perfect sense. I then put in can speak English. Now, obviously none of those animals can speak English, right? But it still has to choose something off that list and it came back with turtle. Can you think why it came back with turtle?

​

Mark A Preston: No.

​

Danny Richman: Because when it looked at all its training data I mean, this is my assumption when it looked at all its training data, it discovered the teenage mutant turtles and saw that of all the animals on this list, the one that was commonly associated with speaking English was a turtle and I felt it would be a good demonstration to help people get their head around what GPT is actually doing. It isn't necessarily designed to give you accurate information. It's designed to give you probable information.

​

Mark A Preston: It's the connection between something and something else, the multiple connections.

​

Danny Richman: And that's all it's doing. So when you put something into GPT, either the training data or your prompt, what it's doing is it's converting all of those words into numbers and then I won't get too technical, but it's looking for something called the cosine similarity. So it's looking for the relationship between all these tokens in order to produce a response. And that's what it's actually working. Now, there are ways of getting GPT to behave a lot more accurately than it did in GPT-4 comes up with much better responses than the previous model. And now you've got things like Bard and you've got Bing's GPT, where what it's doing is verifying information on the web from sources before it then gives you a response to double check that it's giving you accurate information and so that whole area, that whole concern around accuracy is a very short-term problem.

​

Mark A Preston: Is there a way of taking the GPT technology and specifically training it around a subject in a certain industry so it's so accurate, and learns about that specific subject? Say, for instance, it's the dental or optician industry or gardening, whatever it is. Can you specifically use that to train it?

​

Danny Richman: You can do that right now and you can do it at very low cost. I don't know if you've seen; there are all kinds of tools now, where you can upload data and you can give it that data in any form. So it could be a Word document, a PDF, a website, a CSV list, any kind of form of data you want. It then uses something called fine tuning, where it's training the language model on that data that you've uploaded, and that makes sure that it becomes an expert on that data that you've supplied. In fact, I've started doing this now. Sometimes I have long sessions with clients and where we'll have, like, a four-hour strategy meeting, and we're talking about a lot of different things. And so what I do now is it's all done on Zoom. So I get Zoom to transcribe the whole conversation. I then upload that transcription into an AI bot, and then I send a link to that AI bot to the client and said, look, if you got any questions or you can't remember anything about us or what we discussed at that meeting, here's a bot. You can ask questions, and it will pull up that information for you.

​

Mark A Preston: It's just unreal. Wow, this subject really excites me. I've been looking into AI for a number of years now, and it's just the future, I mean, change excites me. I love change, otherwise, I'll get totally bored. For people that are thinking or lost about what to do, especially in the SEO industry. I know nobody can predict the future, but when it comes to AI, what things can they be doing here and now?

​

Danny Richman: So in a very general sense, I would say treat it much like you would with technology more generally. Sometimes I meet people that have been out of the workplace for a number of years, in fact, quite often a lot of women, where they've taken like a career break for several years to raise children, and they're now trying to get back in the workplace and they don't really have much experience of using computers at all. And that's really difficult for them because so many jobs now are so dependent on using technology in some form. And what's the best thing you can do if you're out of step with something is to try and learn as much as you can about it. So I would say as a first step, if you haven't played around with this tech too much, I would say as a first step the easy entry point is, well, probably the easiest entry point is going to be using either Google's or Bing's AI chat function. It's free, it's accessible, it's easy, it's easy to get your head around it. Then I would say step two is to start playing around with chat GPT and really understanding how to get the most from that.

 

Soon you're going to have things like, I mean, I've already got access to them and thankfully, but they'll be publicly available, will be chat GPT plugins, which are really cool. You can do some really interesting stuff with the plugins and then you've also got GPT code interpreter, where you can feed any kind of data into GPT, whether it's a spreadsheet or a website or an image or a video, and then get it to manipulate and do things with that information you put in there. So chat GPT is going to be a great way to play around like a playground with this technology and understand its strengths and weaknesses. But then I would suggest, if you're really quite serious about this and you want to understand how to really utilize it in your career, I would say move away from chat GPT and start looking at the underlying API. It really isn't hard to know how to use an API and then you can start to do some really interesting things because you can get GPT to write you some code that calls the API and then you can start building your own applications and stuff like that. And, yeah, I think anyone that's got to that kind of stage where you can actually now start creating tools and applications and really understand how to develop prompts that deliver excellent results, is going to be very valuable to a lot of companies for quite a long time.

​

Mark A Preston: Wonderful. And just to finish up, is there anything at all we haven't covered that you really think the audience needs to know? Well, apart from obviously yourself and the conversations where to find you and if you actually have time to talk to people, that is.

​

Danny Richman: Well, best place to find me is either on Twitter or LinkedIn. Just put in my name Danny Richman, always happy to chat to people, always try and make time for anyone that wants to discuss anything or need some help or advice. But, yeah, I can't honestly think of too much that we haven't covered. I mean, I would just reaffirm for people to try and overcome any fears or prejudices you may have about this technology. And if you can, try to embrace it is probably the best advice I can give anyone, because it will be coming to your work very soon, if it's not there already. So you might as well get used to it. There's no holding it back now. So, yeah, try and feel the fear and do it anyway. That old adage.

​

Mark A Preston: Wonderful, and on that note, thank you very much for your time, Danny.

​

Danny Richman: My pleasure.