Introduction -
This post is part of our AI6P interview series on “AI, Software, and Wetware”. Our guests share their experiences with using AI, and how they feel about AI using their data and content.
This interview is available as an audio recording (embedded here in the post, and later in our AI6P external podcasts). This post includes the full, human-edited transcript. (If it doesn’t fit in your email client, click here to read the whole post online.)
Note: In this article series, “AI” means artificial intelligence and spans classical statistical methods, data analytics, machine learning, generative AI, and other non-generative AI. See this Glossary and “AI Fundamentals #01: What is Artificial Intelligence?” for reference.
Interview -
Karen: I am delighted to welcome Gunnar Habitz from Australia as my guest today on “AI, Software, and Wetware”. Gunnar, thank you so much for joining me on this interview. Please tell us about yourself, who you are and what you do.
Gunnar: Thank you, Karen. I'm happy to be featured on your show today. So my name is Gunnar. The name is Swedish. Born in Northern Germany, but lived most of my professional life in Switzerland, in Zurich. I work at technology companies; back then it was HP, Hewlett Packard. It was a larger company than today. It was split into many smaller ones. But that's where I learned all about technology and especially moving into future technology type of topics.
That also led me to studying computer science in Germany, at the end of the 90's, with topics like AI without LLMs – so much more theoretical than practical. I think I programmed my first video conferencing systems with touch at the end of the 90's, not too early. So that's a funny thing about that. And I came to Australia at the end of 2016 after a layoff after 16 years, and recognized that I had to reinvent quite a bit.
I moved from a vendor side at HP into an HP partner. Later into other Software As A Service vendors, where I'm today at ActiveCampaign. And I enjoy email marketing, CRM with a lot of AI inside. My role is to work with partners. So some people would refer to it as the middleman, knowing that 75% of world trade is done with partnership.
So I work with marketing agencies who find clients so that they can help make small teams power big business, as we would say. And marketing is so critical these days.
Karen: Yeah, definitely. So thank you for sharing that background. Tell us a little bit about your level of experience with AI and machine learning and analytics. It sounds like you use it professionally, maybe personally. You've also studied the technology.
Gunnar: Yeah, there's quite a bit to it. So back then, when I studied computer science, AI was always kind of there as a future topic to become more practical. I'm also a book author, so I've published 28 books, and I review the books of others. And I remember reviewing a book from an author, Justin Michael from the US together with Tony Hughes from Australia, a book called Tech-Powered Sales, written in 2021 during COVID days. And they talked about all the near future available technology, and they talked about ChatGPT before version three, so quite early. But still, when you read it you think, "Oh, that all is coming very soon". And here we are.
So that's the interesting thing. Working at ActiveCampaign, which has AI in it, like Active Intelligence, it's great to witness what the teams are able to do. In the meantime, things like generative AI, a chat interface compared to the regular software menus, of course, is now very familiar for customers. I love that.
But what I'm sharing here today is, of course, more my personal views and set off from the company side. I mentioned the book topic and my own published books. I remember one thing when ChatGPT came out at the end of November 22. I jumped quite early on that to figure out the opportunities. Because I used to have software like Gong in the past which used AI to do the transcribing, of course. And does all of the analytics when it comes to which part of the sales score is, or discovery, which goes to call to action, which talks to converting, and so on. It was always there, for at least eight to 10 years. But the ChatGPT became more obvious, and I thought the best way what we can do is to do something about it.
And in May of 23, an IT expert published a book about ChatGPT for leaders. And he sold some 30,000 US dollars with this. He published it early 2023. I met him, and I talked to myself, "What if I had the time, for whatever reason, to write a book with AI?" And unfortunately, I faced a layoff. So I said, "Okay, that was a hint. Now let me write one." But it was not good, what ChatGPT produced. I really wanted it to create content, not just use it as a brainstorming part of it. It really was not good. It tends to summarize things. It has this em dash topic, which we don't like in Australia because we don't write em dashes. We have blank en dash blank, instead.
And generative AI back then, also maybe still today, has this tendency to say, "It's not this, but it's that." How humans would never write. So you need to be really careful. And despite that the book under a pen name reached Amazon Best Seller Status, I realized you can do things faster, but you need to be very careful to remain human.
I even have done a concept of a book called "Chat or Cheat" and still have chatorcheat.com, to show more about the other side of it, not the possibility, but also to be dangerous. There are so many things to be really aware of, we as a society, that the humans remain humans. But yeah, I've been involved particularly in the generative AI area and in my corporate role on agentic AI.
Karen: So are there any cases where you have used generative AI or other AI tools in your personal life? Or strictly for professional purposes and you try to ignore it for other purposes?
Gunnar: So in terms of exploration, I used tools, especially generative AI, different ones. Also Claude, for example, Gemini, for the purpose of exploring the opportunities, because it's a chance for everyone. It feels like 30 years ago, something was coming called the worldwide web. I started programming HTML in 93. So therefore getting in early and learning things by doing, I found, is the best way. Instead of putting it to a back burner and waiting until it becomes mainstream and even later where that is, from a job market, pretty bad. Rather figuring out how that would work.
I still take Wikipedia as my favorite knowledge engine because somebody, whoever, I don't mind, compiled something about a topic in decent length. But I enjoy reading, which is there without me asking. Generative AI, from a generative engine optimization point of view, only helps with digging. I cannot ask a question and it gives me as comprehensive as I like with photos and everything.
I think about, like, Sydney ferries. I need to go on the ferry after the call. So what the fleet consists of, and so on, this hyperlink approach of Wikipedia, you can't beat with ChatGPT. You need to dig in it. And if I do the same tomorrow, I get a different answer. So therefore, I still see many traditional ways of knowledge sharing are valid and I use them. But on the other side, it's a good way to be early on and figure out the opportunities.
For example, my new book Happy Habits, I wrote by myself. But in the ideation phase, ChatGPT became a quite good brainstorming buddy. Especially since it can create visuals. Not that they are good or useful, but at least it gets me one thing, one outcome, which I call clarity, and that is good. Sure, I then went to ask people about it because I want humans, not AI agents, to buy the book. But still it gets me on the clarity approach of ah, yes, thanks to this visual aspect of what all happens at the chat before, I come up with the conclusion, which otherwise would not be there.
I've done a funny thing. I got a video from my boss about an event that happened in Melbourne. And he said "Welcome, dear participants. Good to have you here in Melbourne." And then he continued talking for three minutes. And then I thought, “Oh, I ran the event in Sydney and in Singapore.” So I asked him to just record a voice note of all the cities like that, Sydney and Perth and Brisbane and Singapore, so that I could add them to it. But for me, it's too fiddling to do it by myself. And optically, he wouldn't open his mouth too much anyway, so it wouldn't work. So I asked my friend ChatGPT, "Can you please do it?" It said yes, but never delivered. Because a 30 megabyte video in its cache didn't work out. Claude said, "It's unethical to do it. Your boss didn't say Sydney. He said Melbourne. Ask him to record the whole video." And then I realized that AI can also have a more ethical approach to it if programmed well. And I like that.
Karen: Very interesting that Claude declined to fulfill your request and cited an ethical reason!
Gunnar: YYeah, because ChatGPT even, so I use the $20 paid version, is sometimes promising too much and says “I do”, and then doesn't do. That reminds me, my name Gunnar in Australia is often pronounced wrongly as Gonna, knowing that a Gonna is someone who says, "I gonna do" and never does. So therefore, I always need to show that I'm the one, if I announce something, I also execute on it. And otherwise, better not talk in advance about it. But that's how ChatGPT technically worked for me.
I then came on the idea to transition my book, which has four parts about habit building, called Pause-Plan-Practice-Partner, all with a P. And I created a course concept which goes to seven weeks, so it needed to be seven. And then at the end, it summarized it with four. "Hang on. It should be seven." "Oh yeah. Sorry, I forgot your book has four."
We sometimes expect too much, but it is a bit like a human. If you tell a human too long something, confuse them, you also get confusing messages. It tells me I need to be more precise in the way I interact with it. Be more clear.
Karen: I want to go back to something you said earlier. I think I heard you say that you've published 28 books. Is that right?
Gunnar: Yes. Yes.
Karen: Wow.
Gunnar: Under my name.
Karen: And two of them you said you wrote with ChatGPT, but then you decided they weren't good. Can you talk a little bit about that? What prompted you? Was it just an experiment to see what ChatGPT would give you if you had it write a book for you?
Gunnar: Yeah, it was actually, indeed, kind of an experiment. It's not included in the 28. They're all under my name. Back then I thought, "What would ChatGPT know that I can write something about?" And it was a book about affiliate marketing using the power of ChatGPT. It was early 2023. I knew how to create books. It was just after I published my own self- published business book about LinkedIn, "Connect and Act". So I had the infrastructure of producing books easily. I knew how to do that on Amazon, Kindle, all of this kind of thing was easy. But what I wanted is to find out how I can write a book.
I was sitting there. I was thinking for one hour, by myself, what it could be. And I thought, being an affiliate marketing specialist, I had an agency for that in the past for email marketing and landing pages. So I knew many of these programs. And also I faced a layoff at one company. So I place it in a story sitting in the middle of the US, in the state of Arkansas, where the gentleman would lose his job at a private post, competing distribution company, and need to find out what else is out there, how he can make money using AI, but not only AI. And I asked that it should have been a factual type of book, like research. I wanted to find out how much ChatGPT can help me to get on research topics. And I thought the best way to put this in would be in the context of humans. I asked it, "Could you also create a story with humans in it?" And then it invented a couple. So I gave them a daughter too, but I kept the original names of the couple. It was quite funny.
And then I figured out where in Arkansas I've been that I can place it, like a regional story, and whether people would go and travel and get feedback on it. It was for me more exploration, how fast it works. It was 16 hours of prompt engineering, writing. One hour of thinking makes it to 17. Four hours of reading makes it 21. Two hours, editing, uploading, and one hour photo and upload. So 24 hours work. But not my name, because the quality of the type is not what I write. And back then you couldn't give too much of your own writing to it, that it would do something out of it. It was with 3.5, not 4.
Karen: I'm really curious. One thing that we hear a lot, especially with the way that AI is trained, it tends to kind of average out different scenarios and mashes them up. And when it comes to regional details and specifics and geographies and accents and context, that's where it tends to fall down. I've interviewed a few people from Australia and I've heard them talking about difficulties. For instance, trying to get an Australian themed background, and it would give her these fairy floss trees, which was not at all representative, maybe gum trees would've been better, right? But that's not what the ChatGPT gave her. When it comes to Arkansas, there's a very specific southern US culture, even specific to Arkansas. And I am wondering if it really had any chance of getting that right. You would have to know enough about Arkansas to be able to verify it, I would think.
Gunnar: Exactly. An extensive trip beforehand helped me to put it. It was my idea. I didn't even tell ChatGPT that I would place that. These are the parts that I wrote by myself technically, like the introduction, into the five chapters. So that I really placed that, where they would go and which restaurant they would sit. I took my old travel photos and figured out, "Ah, yeah, that would be in this restaurant, and that person would maybe work there, and then you take the light rail over there" and really discovered a trip that I've had not too long ago. So that was a joyful part of it.
And I did not want to place it in New York or L A. That would be where this type of normal people situation would be different. Because you think in New York or San Francisco as a startup scene, you would have more modern chances, compared to in the middle of the country. And therefore I picked Arkansas, having been there, enjoyed going there. But that was my part to it, with research from people that I also met there and are validated with them.
That's one thing that ChatGPT actually is quite bad at. So when I wrote a book last year, my mom said, she's in Germany, "Oh, I don't understand English well enough. Can you translate it?” And no, don't use ChatGPT for translating, because it tends to summarize things and I complain. I need to have the same type of thing. It doesn't matter if you know where it's heading to, to summarize the sentence.
I used a German AI product called DeepL. Not sure if you know that, deepl.com. It is for academic research translation that is actually very good. It has a very good reputation. And what it can do, when I have my book, let's say in Word, everything on it with all of the style guides and everything, I just build. There's a bar in DeepL on the right hand side. I click on a paragraph. As I translate with DeepL, it shows me. I say, "Okay, good enough." I changed it a little bit. Clicked on a button and it copied it in where it should be. It pasted it where it should be in the right formatting.
That is a massive game changer to have this type of integration with Microsoft. And the team at DeepL has done a good job. It's German. Maybe not everyone knows it, but it's one of the tools which in academia are actually used to have a starting point, before you give something there to a real editor. But still, it's much better than I would be able to do.
Karen: So it was the focus of the tool translation specifically. I remember when I was in high school and trying to figure out what I wanted to study, I wanted to become an engineer and my uncles told me, "You should learn German. 'cause you're going to find there are papers in German." Okay, well, I'll see if I can learn enough German to be able to read some papers. My German never got good enough that I could understand the full paper myself. But it would be awesome to have a tool that would actually let me understand the paper as the person wrote it and know that it was fairly accurate. That would be quite a game changer.
Gunnar: That actually is also the purpose. Academic text would start with the proper summary on it. And then have all of the proof coming up there. In social media, I would say it starts with a TL;DR. You only read this, and everything else you don't need. I like that because when I started to work as a journalist while being a student, the editor always said, "Write in a way that I would like your whole text. But I might cut the last two paragraphs because I got some advertising in at the last minute. But your article, you need to still work." So the abstract, like an academic text, type of introduction should be there, that everything is there, but if some paragraph at the end with further proof is not there, it should still work. And that's where this DeepL AI tool works actually quite well. There's even, I think, a one month long free trial, which was good enough for one project. But in the meantime, I have a paid version because I used it.
And the funny thing only, I grew up with the German language. I use English for 25 years or whatever, but I translated my book about leadership. And there were stories with digital marketing. But in Germany, where I hadn't been living for years, would they use terms like landing page as we call it, landing page or website? Or do they translate everything? How would I know that? And that's where AI tools can help, but a human must be there to really validate that and say, "Oh yeah, no, we don't call it a landing site. We call it a landing page." That's fine.
Karen: Were you able to generate a copy for your mother, then, to read in German?
Gunnar: Yeah, yeah. Not only this, I put it on Amazon.
Karen: So you're selling the German translation! That's very good.
Gunnar: Yeah, yeah, yeah. It actually sold a few more books on Amazon than the English one! That's quite funny. There's one aspect to writing in different languages. So in addition to my books, I also wrote 450 articles in print, most of them in German back in the day. And I know one thing. Before the blogging, I have a tracker that tells me every piece I wrote and published. How many words, how many characters? Because 20 plus books were travel books, like Lonely Planet type of publishers, internationally speaking. But for German ones, what for me was important to write on format. And in the middle is an insert of two pages with another layout because it will use a topic in it that needs to be two pages. Cannot be two and a half. So I used to write in formatted characters, not in words as a novel writer would do.
So then I know one thing for sure in English, an average English word is six characters including the blank. In German, it's seven, exactly one longer. So my book from 88 pages turned into 96 pages in German because it's just longer. Therefore, I like English, because you can express something shorter, more precise without the need for long explanations. And that makes all the difference.
Karen: That's very interesting to know. Yeah, it makes sense. One thing I wonder, you mentioned that you've written 28 books and 450 articles? That's quite a lot of output! One concern that comes up nowadays is where AI and machine learning systems get the data that they use. In a lot of cases, we've heard that they've scraped books, articles. If you remember hearing the story about LibGen, that people were able to go in and look up and see if their works were used to train AI without permission. So I'm curious if you had any experience with that, if you ever looked yours up? And how you feel about what they call the “three Cs” of Consent, Credit, and Compensation, and whether people should be entitled to have that, or if it's okay for these companies to just take the data that is, as they call it, publicly available, even though it's not public domain.
Gunnar: First of all, as a book author, I can write my own book, but I can publish works that are in the public domain. Think about even operas. Or I'm not sure if I'm correct about Shakespeare. But let's say other writers where the copyright expired a long time ago, for whatever reason. There's a key difference between what can I do and what should I do. So it's a question of respect.
A friend of mine who was in the musical scene in Europe, he one day complained and said, Hey, there's this singer Sarah Brightman, who we all know and love. and she used a song of an opera where I now earned the copyright. And he asked himself, should he be really disappointed, go the legal route? And he decided to be proud that she took this page from a Czech composer Antonín Dvořák with the new lyrics, kind of, despite she didn't have the copyright. And he told the story based on gratitude instead of legal threats. Much better. I like that.
But in general, of course, as a writer, there's a tendency to say, for 90%, I would be disappointed that AI should take my words for learning. Citing properly would be fine, but typically it wouldn't. And that, of course, is the negative side of things. For 10%, it tells me, "Hmm, the more I write, the more I can influence on a particular topic." If I'm kind of a topic owner, my view to things, or my research, can make its way into GEO.
But in general, I'm not too happy, but I can't change anything about it. So I know that happens. I know that a lot of work is taken to use it. So therefore, from the receiving side, we need to take everything with big care on the genAI side. Because when it says something, we shouldn't take it for granted. And too many people if they do that, they said, "ChatGPT said it." I said, "Well, what about ‘Wikipedia said it’?" Even there, people have doubts where it comes from. I can change a Wikipedia page, and I've done, all over the place.
So I'm critical in general about it. But what I can do as a single person is maybe a bit limited. If I can influence something, the more I talk about a topic, the more I can be cited that GEO can help. It's also a way particular for those who are, let's say, a specialist on one particular topic. So that of course is a possibility.
But yeah, there's a lot of things to be careful about the usage of AI, and that's one big one. I have another one as well. That has to do with exploiting resources on earth for laziness. Back in the day, I would go physically somewhere, exercising to buy a DVD, put it into a drive, watch it because I own it. Maybe I could even borrow a DVD. But it would not require the power of network data to be transmitted for streaming. Where today we are lazy. So how much energy is used for this laziness? How much, if I only say "Thank you, ChatGPT", it costs billions to OpenAI to just recognize this tiny fraction of a second of energy consumption and resource exploitation on earth because people are friendly or lazy or something like this. So that is worthwhile to be critical about.
Karen: You mentioned GEO. For the audience members who maybe aren't as familiar with that, you mean generative engine optimization?
Gunnar: That's it. I coined one term as well, called MEO with M, message engine optimization. That was maybe last year where I said the most underrated search engine on the planet is LinkedIn messaging. And that goes in the same way as SEO and GEO. I need to plant something so that it can be found. SEO is only there so that Google, before AI terms, would show a list of those people who don't have the best website for the request, but who worked SEO the best way. So it needs to be planted.
GEO works best for generative engine optimization if the content that I have can be cited by different AI tools so that it'll pop up now. But it's a different way of thinking than SEO. So if I would connect with you on LinkedIn, or connect, you accept for whatever reason you believe it makes sense , you don't know why I reach out. If I then don't do anything with it, and of course I don't pitch straight away, then there's just a connection which is meaningless. So therefore the better, I start this by putting a person's message with two hooks at least, and the hooks are built out of future search terms that are put into legal methods, I can find again the person. So it's the same approach of planting first, knowing what will come up. That's the idea.
Karen: Yep. Very interesting. Have all your books, your 28 books, been self- published?
Gunnar: No, 22 I think are books from professional publishers. I think three of them have the title "1000 Places to See Before You Die", which is a kind of a franchise type of thing. The most-gifted unread book series in the world because it's good to give it to someone, but it's not really read. And my publisher in Germany brought this type of license to use this occasionally. So there, it's more writing books as labor. Then, half year after submitting the whole professional thing, and printing them somewhere overseas, we shipped, and it's available in bookshops. So every bookshop in German speaking countries would have at least five, if not more of the books that I wrote, or contributed to.
But I started in 2023 with self-published books. One fairytale. Two LinkedIn related books. And now a book about habits because it's my last name!
Karen: Let's talk about your Happy Habits book.
Gunnar: Yeah, so I was sitting on Easter Saturday at a bench in Sydney, a nice atmosphere outside. And I thought it's a good time to read a book. And that was from an American author called J.R. Heimbigner. He's promoting the idea of writing mini books, something like 60 to 100 pages. And this book was about discipline. And coming from a German-speaking background, I think that is something that people there would have, to read something and do something really in a disciplined type of approach.
And after 25% in, I thought, "Hang on, I can actually write a book by myself." On one particular topic. And then I reached out on the same bench sitting there to my good brainstorming buddy called ChatGPT and thought, "Hey, what about the idea to do a book in the series? Have a look at what he's writing." And I always wanted to write about habits, because I carried it in my last name. And I wanted to add something new to the discussion in addition to the famous books like Atomic Habits, Tiny Habits, and so on. And then after two hours prompting back and forth, and also chatting with two people, I came up with the idea to not only call it Happy Habits, but also to have a good structure on it. And what I liked was the clarity that I got from it, and I put it into four elements because it's my lucky number. I'm born on the fourth.
I talked about the layoff topics three times in the book. Quite a bit about planning things, about how to practice approach towards habits shaping, and finding and improving, like habits stacking for example. And then also about partnership. I work in partnership, but also strongly believe that if you try to do something in isolation by yourself, you can only go that long. But if you have accountability bodies, and I have a lot of personal stories to this, then it goes, of course, much better.
And that was how I phrased everything together. Actually AI helped me to shortcut the ideation process. I already must admit that. And then I thought, being in April. And it'll be my book number 28. And it had to do with what I learned from, particularly, my father who wrote for 20-odd years opera critiques with an amazing precision of repetitive and forming habits of doing things. And I thought, okay, he would have his birth on the 28th of July. So I set this stone, the 28th of July would be the launch for my book number 28, which is related to him. And it was his dream to move to Australia, so it fit all together. And I read the signs for that and I only had to back-engineer to figure out, when do I need to finish? And put the milestones in between? And I got it out.
Karen: In time for your father's birthday on the 28th. That's amazing.
Gunnar: Yeah, exactly. Yeah. But he passed away four years ago and I think he would've liked that. And my mom has been on the call as well. And she's now writing books also.
Karen: Very good. Yes. Is your mom also using AI tools in any way for writing her books?
Gunnar: Yes and no. Her first book took her one and a half years where I coached her. The book was about Parkinson's, because my dad passed away with it. And she wrote in the book about the role of a caring partner, not a medical textbook, but more how can we prevent, not the disease, but the speed of you get it? So then when certain signs are there, what can we do against it, when it comes to nutrition, when it comes to exercise, and brain power. How can you do things? So therefore, I try to avoid many things that my father then ran into, also thanks to the help of her book.
In January, the German version came out. As soon as she finished, then she realized her dream was to write a childhood memories book for her family, when she would have around her birthday in April. And she went into a club of seniors learning about ChatGPT. And then she had dialogues with ChatGPT. And out of this came three dozen little stories that talk about childhood memories. I turned that then into a readable book. But it was actually pretty good, what the output has been. Short stories, one page to four pages. That was good. So she used it for the second book, and it took then six weeks instead of one and a half years.
Karen: Yeah. That's really good to hear that it's helping people like your mom to be able to get their voices out there, when otherwise she probably never would've been heard. Her stories never would've been heard.
Gunnar: Yes. Just, we need to be mindful to have human editors so that it doesn't read too robotic.
Karen: Yeah. Makes sense. So I'm curious, have you ever tried any AI-based tools other than for words? Like for generating images. You mentioned something about a symbol for your book. But for images or for music, anything like that?
Gunnar: Yeah, so you recognize a lot about the generative AI. I mentioned Gong earlier. Gong is a software that companies use to record calls. I think they started in 2016, and know the head of sales from back in the days. You would not have AI tools in that sense. And today when you have Fathom and Otter and Read.AI and all of the little tools that came up, left, right, and center.
Gong is one of these big, massive tools that existed for a long time. The license can cost six digits per year for enterprises because of the analytics. Massive what you can get out of it. So normally people would not have that. So that was the first part of how I got into AI, and then generative AI, to use it.
But funny thing, early this year, I thought about adding a podcast for myself. I call it Behind The Book. I never started it, but will be on my Substack one day. And for a podcast, you need to have some kind of music, some jingle to get started. So I reached out to Suno, which is an AI music creation app. I published my first album of piano songs that you can play in a cafe because I love cafes. And back then, having the paid version, I could create things where legally I could add my copyright to.
And I've done the reverse part of it. I went to a club that turned 150 years old. I composed a song for that. No, I prompt-engineered a song for that. I should not say compose. It took Suno two minutes, out of the clear instruction when it comes to the text. One should be there for the target audience to record professionally this song. And then I took two other gentlemen who were in the same club as well. 70 people in the room. I had the text printed to give it out to someone. I gave a tiny singing lesson of the chorus that everyone would sing in with all of us. I played it so loud that nobody could hear me because I can't sing myself, and the two other people began singing this, in the beautiful happening. And the president of the club at the end, when he left the room, he looked at me and said, "Maestro". I think that was quite, quite nice as an Italian word for master, in a musical scene. But that was it. "You compose it?" What can I say? I didn't compose it. They recognize normal folk songs, even hundreds of years ago. They work based on predictability of melodies, of chord structures, of lyrics, and that's what generative AI can do. It's not revolutionary, it's just predictive. But the effect was amazing. So Suno was quite fun on that one.
At work, I'm happy to use agentic AI in the sense of, “Should I create a whole email campaign, a series of seven emails by myself? Or should I use generative AI to craft the copy of a seven part email series? But why not then have the agentic AI to write the whole thing? Why should I bother?” And that is where I see many trends of AI are going, to not only generative AI, but getting things done. Why should I care about it? And that's an interesting part where AI is going. I enjoy that.
Karen: You've mentioned working with the company ActiveCampaign that creates AI tools. Can you say anything about where they get the data that they use for their campaigns?
Gunnar: Yeah, there's one big distinction between generative AI and what's up here. Because if I'm a customer using ActiveCampaign, I would love to have certain, let's say analytics, I can read reports, everything.
But as we all know, in the meantime, ChatGPT type of prompting, there is a part of ActiveCampaign which allows to prompt the analytics around it, in the sense of, "Can you compare please my campaign performance in the second quarter together with the typical industry performance? Is it actually good what I've been doing?"
And then it grabs public available data when it comes to industry sources, but with my own data. That means the AI takes that particular client's data out of ActiveCampaign, of course, ignoring any other client's data because of data sovereignty. So we could not use OpenAI, for example, all of those where every data is shared in public. So therefore they have a special way of including this, based on confidentiality of data. That's critical.
Karen: Great to hear that. When thinking about the ways that our personal data and content has been used by AI-based tools and systems, is there anything that you can think of where your personal information has been used, and how you felt about that?
Gunnar: Yes. So it reminds me of one situation when I had to extend my passports. I'm German and I'm Swiss, and I went pretty much at the same time to both consulates in Sydney. The German one was a bit old school, a lot of paperwork and a birth certificate, like maybe a hundred years ago. And sent me back because unfortunately I took the copy of the birth certificate instead of the original, which is anyway just a paper that they sent.
But then I came to the Swiss one where they said, "Oh, we don't need any paperwork because we know you already. We have you in the system. We see it's you. Please go to that machine." And they take the photo again and put your fingerprints. And I thought, "I've done them so long ago. My fingerprints did not change." And the lady said, "You might remember that as we have our referendum four times a year, once we voted against storing biomedical data, even when the passport is produced. So we cannot use it anymore. We need to collect it again."
Instead of seeing this as a burden to put the fingers on, I actually felt very happy that when we vote something as a democratic process of our referendum, that the government sticks to this and requests it again. It's like when a company says "You can only log in here with two factor authentication." I never see it as a burden, but with pleasure. Yes, that company forces better security. Switzerland forces me to put the fingers again on it, and I know they don't store that. And then, in eight years, I need to do the same thing again, happily. I like that.
Karen: Yeah, that's great to hear, especially because biometric data is so dangerous. Because we really can't change it if it gets hacked. It's one thing to change our phone number or email address, but like you said, our fingerprints don't change.
Gunnar: Exactly. I know in the past, before AI, sometimes people are grabbing data and stealing this. So one day I wondered, "How come Google knows my age?" Then I figured out it cites a source out of an online magazine. And then that online magazine, it consists of many people who contribute to it. And one person contributed and shared with them a CV that I had from 2002, back then when I'd been at HP in Switzerland. And a CV I think is too much private info. It includes a birthday, because without a birthday and photo with German speaking countries, back then it would be disqualified. Including both in English speaking countries disqualified as well. But therefore it was there. So then I reached out to that online magazine. And they really deleted everything from that particular content creator, because I had the proof that there was no consent for that type of information. But Google still shows that it knows it. What can you do?
It's tricky to really get down into deleting something. When I worked at Hootsuite in social media, there were tools available from companies like Proofpoint that I partnered with, where it's possible to delete something off Twitter back in the days, or Facebook, or whatever. It's possible, but a big effort. But try as a small business owner, when you have a shop, it's connected to Google reviews and then one of the Google Review experiences, one person was maybe not good, but written in a nasty way. That can kill business. It's very hard to get rid of that.
If you take this further with AI, that's a very tricky point, how to figure out that it's my own content used, and what to do with that. So there might be experts who have more chances to do something against, right? Agentic AI tools, at least, I can point to that activity work and say, "Hey, your content has been used here or there. Let's go after that."
Karen: Yeah, and obviously stealing data has been happening for many years. People have been plagiarizing articles on LinkedIn for years, and all of this precedes AI. The one thing that people point out with AI, though, is that once your data gets pulled into a training set, for a large language model, there's almost no chance of getting it out of there.
Even if you can correct it at the data broker source, or get it deleted, it's really never going to come out of those AI models. And it kind of defeats, to some extent, the right to be forgotten, which some countries have. We don't have it in the US, of course. But it certainly makes it a lot harder.
Gunnar: Yeah, I heard about that, right? Yeah, that's a tricky thing. Whatever is out there, even back then, an ordinary way of the internet, and with AI further, you cannot do something when it's there. You cannot delete it off the source. Hundreds of years ago, it was enough to burn books to ensure that the writing in that book will not be, let's say, published further. As if it was eradicated, as if it never existed. No way today to make that happen.
Karen: I'm glad to hear, though, that the magazine was able to act in response to your objection in pointing out that they didn't have your consent to publish that. I'm glad they took action on it. So one of the reasons I was wondering earlier about if you were publishing your own books is that there are a lot of lawsuits. I think 42, the last time I checked, in the US alone on publishers, companies like Disney and book publishers who are suing. So it's one thing for, I think, the bigger companies, they have some leverage. Smaller publishers or self-publishers really don't have that kind of leverage, but hopefully if they can set some precedents from the big suits, then maybe that will help to protect the smaller publishers as well.
Gunnar: Truly it's because, at the end, the whole publishing industry requires maybe three main areas. Someone who writes it. Someone who reads it. And someone who joins the two together. And we all do it in a good faith and to bring the knowledge of humanity further. So if 80% do this well, or maybe even 95% do it well, those 5% can raise a perception that prevents many people from sharing good things. And that is a sad story.
But humanity always means we need to live in balance. And if we believe in balance of good and evil, we need to accept as much evil as good. And so far, the good part for me is more than the bad side of it.
Karen: Yeah, talk about people not sharing things –- we've heard some talk about that from musicians saying, "Well, if they're just going to steal my music, then I'm not going to publish it online. I'm only going to perform live." Or you see people who are pulling out of social media sites. People that I know have left Facebook saying, "I'm not going to share anything anymore there. I'm going to send my family pictures to my folks through Signal or something like that. Or just email them." They're pulling away. And it would be a shame for society to completely lose that, the ability that we all have to enjoy music and writing that is free and published online, if people decide that it costs them too much in terms of their livelihoods to have it that open. So there are definitely consequences on that side.
But on the other side, having these tools available for people that, for instance, maybe have disabilities and they use those tools to help them overcome it, or to write in a second language or such, those are obviously benefits that help ideas spread and to reach people that maybe need to hear them. So it's a balance, like you said.
Gunnar: It truly is always a question of “Should I look into the negativity side of it first or the positivity?” So if, for example, let's assume I write something and I'm coining a new term, something that really brings humanity forward? Not the term message engine optimization. But even that, even if I take this and always talk about it and on, I did already on LinkedIn and Substack. And if that is quoted in the future somewhere else thanks to LLMs, not caring about copyright too much, then I can try to force against it and put all my energy into fighting against it. Or I can recognize there's a chance that my thinking gets broader acceptance. I can also try to utilize the power of that.
But the bad thing is if I am a content creator of negativity, think about what free speech allows, but should not. Then of course, that's the downside of it. And for us as humans who are reading, if you take our friends of tech tools who are using content that is somebody to do something with it, to analyze or to phrase it or whatever, we all need to be mindful that the sources can be not that trustworthy.
Karen: Yeah, exactly. And if people don't know enough about the subject to check, to know that it's accurate and they share it, then that can obviously be very harmful to people, where the wrong information is shared about them, or it can actually cause harm. Like you said, there's a balance between using the tools and, in some cases, people just ignore the fact that the tools have caused harm or could cause harm, or not being aware of what the risks are. That's actually my first book due to come out next month. And it's about the key risks and understanding what the ethical concerns are, the impact on power usage and other aspects of AI. And it's not saying because of these harms that no one should use them, it's just that we should be aware of both, and we should be trying to mitigate the harms, the evils, in parallel with trying to find good uses for the tools. So it's looking for that balance.
Gunnar: So back in the days when content has been plagiarized, because it was technically possible, that people would quote it and put it as their own stuff. Think about even newspapers. The key question when I'm the originator of an idea is maybe twofold.
One, I would be happy that my ideas and thoughts get further. And even think about Albert Einstein. He didn't have the power to do a lot of practical tests of his research ideas and theory, so that many people took the ideas and tried to do something about that. But that's not a bad way to get some content out there further. Let's say even in academia, you would have peer to peer reading before something is really published to get the broader experts on the field to go on it. That part I like.
But when something has been shared further, as a contract writer, I would also like to leverage, benefit maybe financially, maybe speaking opportunities or whatever. And that's the tricky thing if content is taken without informing the source. Then maybe everyone knows to talk about something, but the real creator of that who didn't have a chance to get a patent of a topic or a trademark or copyright is then left alone and poor, instead of recognized for what they put into it. And that for me is one of the big dangers to it.
Karen: Yeah, definitely. Of the three Cs, the one people seem most concerned about is credit. Credit doesn't necessarily have to cost anything, but give people credit for their ideas, even if you're taking them and spreading them further. And the tools have to make it feasible for people to know the credit and where it came from.
Gunnar: That's true. And therefore, in my books, I don't just say, "Oh, that's my book, that's all my ideas." In my business book, I always add the voices of other experts, because I strongly believe that you curate the voice of others who maybe are more clever in some areas of that. I have the four areas of my Happy Habits book. I added four experts on that. And also the Foreword, carefully chosen so that I know that they have that excellent level of information that they can share with that. And I had them in my launch call as well. Two of those. That was really good.
And I even crowdsourced the cover design of the book to the community on Substack, on LinkedIn. And the funny thing about that book, in Happy Habits, there is an image on it. The image actually came from ChatGPT, in the meantime. So I'm not using Midjourney anymore, because ChatGPT is good enough for getting an idea. But it was the first time to really use it, and print it in print quality. And therefore when I ask, “Please use a teal color”, it has done that. But therefore, based on what came out of this, I had to take the exact color and the exact image for it, and it looks good. I'm very happy that I thought about happy.
I also added a smiley to it, for two reasons. Number one, it adds a more fun element to it. But also recognize, in English, happy and habit – happy has five letters, habits has six. The I is short. It's not very wide. So I had kind of a half normal letter size and the Y is ending on a line. So I had this space. But a smiley would be good. And I also put it in yellow. And I really liked that.
Karen: I have a personal affinity for smiley faces!
Gunnar: What a surprise!
Karen: I enjoyed seeing it on your cover!
Gunnar: Yeah. And sometimes the name brings something on. Like in my case, habits was long term a dream to do something with that. So therefore, finally, it's there. And very happy to now even create a coaching program and online course based on that one. So it gets bigger than just a book that I wrote.
Karen: That sounds great. So we'll put the link to your book into the interview article and publish it so that our readers can go and find it. I went and grabbed my copy right away.
Gunnar: Thank you.
Karen: As soon as I saw the post from Fleur in the Substack group. That was great to see. I'm looking forward to going through the knowledge that you're sharing there and making good use of it. So thanks very much.
I want to respect your time. I know you've got a deadline here. Is there anything else that you would like to add?
Gunnar: I thank you for the invitation to talk about the usage of AI. I think for all of us, it gives us great chances. It gets us constraints we should be aware of. They're changing all the time. Whatever we thought doesn't work with AI half a year ago today can be much better. Whoever says AI will take our jobs, think about 30 years ago, there were no digital marketing jobs. There were no SEO experts, not even websites people around. So life is always changing on this one.
I'm happy to work at a company where we all are learning and training on AI every week. And to be well advanced on that one, very happy for this. And I only believe that those who may not be working in an AI company still can be familiarized with the chances, well aware of the risk. And reading content from experts like yourself, that can be very helpful.
Karen: Great. Well, thank you so much for making time for me in your morning. I appreciate it, Gunnar. Thanks so much and have a great day, and I'll look forward to catching up with you in Substack soon.
Gunnar: Thank you. Thank you, Karen.
Interview References and Links
Gunnar Habitz personal website
Connect And Act book by Gunnar Habitz
Happy Habits book by Gunnar Habitz
Gunnar Habitz on LinkedIn
Gunnar Habitz on Medium
Gunnar Habitz on Substack ()
About this interview series and newsletter
This post is part of our AI6P interview series on “AI, Software, and Wetware”. It showcases how real people around the world are using their wetware (brains and human intelligence) with AI-based software tools, or are being affected by AI.
And we’re all being affected by AI nowadays in our daily lives, perhaps more than we realize. For some examples, see post “But I Don’t Use AI”:
We want to hear from a diverse pool of people worldwide in a variety of roles. (No technical experience with AI is required.) If you’re interested in being a featured interview guest, anonymous or with credit, please check out our guest FAQ and get in touch!
6 'P's in AI Pods (AI6P) is a 100% reader-supported publication. (No ads, no affiliate links, no paywalls on new posts). All new posts are FREE to read and listen to. To automatically receive new AI6P posts and support our work, consider becoming a subscriber (it’s free)!
Series Credits and References
Disclaimer: This content is for informational purposes only and does not and should not be considered professional advice. Information is believed to be current at the time of publication but may become outdated. Please verify details before relying on it.
All content, downloads, and services provided through 6 'P's in AI Pods (AI6P) publication are subject to the Publisher Terms available here. By using this content you agree to the Publisher Terms.
Audio Sound Effect from Pixabay
Microphone photo by Michal Czyz on Unsplash (contact Michal Czyz on LinkedIn)
Credit to CIPRI (Cultural Intellectual Property Rights Initiative®) for their “3Cs' Rule: Consent. Credit. Compensation©.”
Credit to for the “Created With Human Intelligence” badge we use to reflect our commitment that content in these interviews will be human-created:
If you enjoyed this interview, my guest and I would love to have your support via a heart, share, restack, or Note! (One-time tips or voluntary donations via paid subscription are always welcome and appreciated, too 😊)
