Human Intelligence Podcast
Episodes as they happen...
In this episode
Andy and Mark have known each other for three years, speak every month and work together, yet they’ve never met in person. So when AI suggested the topic “digital bonds are shallower than real-life ones,” they had a vested interest in the answer. Mark went in firmly believing face-to-face relationships always win. By the end of the episode, he’s not so sure.
Read more…
They explore why friendships built online often have more regular contact than decades-old school friendships, why you only ever know the “work version” of your colleagues, and whether people are more honest behind a screen. Andy shares why his catch-ups with former colleagues always slip back into the past tense, and why numbers at his local networking event have dropped since offices reopened. They also talk about the exhausting cost of putting on a front at work, and why showing up repeatedly, online or in person, is what really turns contacts into proper relationships.
Key takeaways
Frequency matters more than format. Relationships deepen through regular contact, whether that’s online or in person.
Friendships formed digitally often have more touchpoints than old in-person ones, so you know more about what’s happening in each other’s lives now.
Relationships without ongoing contact tend to get stuck in the past, with every catch-up returning to old stories.
Moving from comments to a real conversation, even over video, is the moment an online connection deepens.
In-person events like networking groups only pay off when you go repeatedly. One visit rarely builds anything.
The hosts’ verdict: neither format wins outright. A healthy balance of digital and in-person contact works best, especially when everyone is busy.
Chapters
00:00 Welcome back, and a topic picked by AI
00:31 The claim: digital bonds are shallower
02:24 Three years of friendship, and we’ve never met
04:40 Why old school friends drift and online friends don’t
06:28 What changes when you actually start talking
08:08 Where’s the line between shallow and deep?
09:22 What lockdown Zoom calls taught us
10:59 You only ever know the work version of people
12:34 Are people more themselves behind a screen?
14:17 A quick word on swearing
14:52 Old colleagues and living in the past tense
16:56 Dialling yourself up and down
19:02 Scottish accents versus subtitle software
19:38 Why networking event numbers are dropping
23:08 Paying to build a habit
25:06 Life-changing conversations take repeated effort
26:06 The exhausting “work you”
28:48 The verdict: a healthy balance of both
30:20 Over to you: a pint, a screen or both?
Full transcript
Welcome back, and a topic picked by AI
00:00Andy: Hello everyone, and welcome to episode two of the Human Intelligent Podcast. You know the crack by now. Mark and I discuss various topics in the world of humans versus AI, or humans connected to AI. We’re not AI haters in any way, but there are lots of topics to discuss. And today, we asked AI to list a few options we could talk about. Hence, non-AI haters.
Mark: It should be called Hypocritical Intelligence.
The claim: digital bonds are shallower
00:31Andy: It should be, really. I named it when I was probably a bit more of an AI hater than I am now. One of the topics that caught my eye was digital relationships versus ones formed in person, and the theory that digital bonds are shallower than bonds formed in real life. Mark and I are middle-aged blokes,
Andy: and I think a lot of people our sort of age, maybe those who aren’t in the digital space, would default to: it’s definitely better to meet people in real life. Digital bonds are more shallow. You don’t really get to know someone unless it’s in person. But the research doesn’t support that. What it’s saying, and I want to get Mark’s take on this in a second, is that digital bonds have
01:27Andy: more touch points. For instance, Mark and I have known each other for three years. We speak every month and support each other’s social media content. I know a lot more about what’s going on in Mark’s life than I do about people I’ve known for 40 years who I don’t speak to for six months at a time, because our relationship was formed in the physical world, at school, and we’re just not in that digital mindset of
Andy: checking in and speaking all the time. So I might not know what’s going on with them for a year at a time. Whereas Mark and I are in contact all the time, in lots of different spaces. That’s what the research is saying, as a challenge to the idea that digital relationships are shallower than physical ones. So Mark, what’s your take, based on what you know and how long we’ve known each other?
Three years of friendship, and we’ve never met
02:24 – Mark: I’d absolutely have been in that first camp: no, in-person relationships are definitely stronger. But as you say, we’ve known each other for three years, and we’ve never met, and we might never meet, just because of geography. You’re at one end of the country, I’m at the other, and it’s an absolute bollock to get from one end to the other. But yeah, that’s a fascinating point.
02:52 – Mark: You used the T word. I’m going to have to use the T word. There are more touch points. Sorry, I’ve just been a little bit sick in my mouth.
Andy: We’ve tried so hard to come up with another term for that. We just keep coming back to it.
Mark: It’s not good. I’ll keep trying. And I’m not going to say “journey.” I’ve just said journey, but not in that context.
03:17 – Mark: There are so many more touchpoints. It’s easier and more convenient to check in with people digitally. It’s like the old idea that you spend more time with the people you work with than anyone else, because you’re with them
Mark: eight hours a day. There’s that constant contact, that checking in with a purpose. Whereas with in-person friends, I’ve got to go all the way into town, so it’s got to be worthwhile. You don’t see your in-person friends as much as your digital ones. Plenty of people’s friends will be on the digital platforms too. My friendship group is constantly
04:03Mark: on WhatsApp. I don’t do WhatsApp. I’m firm on that, and I will die on that hill. Much like Trevor and Simon’s “we don’t do duvets,” I don’t do WhatsApp. So they message Mrs T and she passes it on to me. But until you said that, Andy, I hadn’t thought about it: yes, you do get to know someone more. And it doesn’t have to be face to face, like we’re doing across video.
Mark: Just commenting, just messaging. All forms of touching base. God, that’s even worse than touch point.
Why old school friends drift, and online friends don’t
04:40Andy: Do you know what, mate, I think this is where the line gets a bit blurry between digital and real life. Most of my older friends from school and college aren’t social media creators. They don’t spend much time on social, and they’re certainly not people who’ll comment on and support social media content. But because of the world we’re in, we’ve met a lot of people over the years who do.
05:07Andy: We’ve become friendly with those people, and because of all that contact throughout the week, we feel more connected to them. We support them, they support us, and a friendship bond forms.
Andy: Coming back to my point: even though my school friends would still be classed as my oldest friends, because they’re not creators, we’ve massively grown apart. We might see each other once or twice a year, but we don’t have the kind of relationship where we’re checking in all the time. Whereas
05:41Andy: relationships formed digitally can sometimes evolve into in-person ones too. There are lots of people I’ve met online through my work who I’ve then met in person. Obviously there’s the Lego crew, who we’d love you to be part of, but again, geographically it’s not possible.
Mark: You’d have to carry me in on a tablet or something.
Andy: That’s a relationship that started online and has evolved into in person. We’re now very, very good friends.
06:10Andy: And I’m probably closer to them. I know more about what’s going on in their lives than I do about people I’ve known for 40 years. Hang on, that’s what I was going to say. I’ve nailed it.
What changes when you actually start talking
06:28Andy: Where I was going with that is: you could support someone online through their social media content for years, following each other, a mutual situation, and your only contact is comments. I think that relationship takes a massive leap forward when you start doing this sort of thing.
06:53Andy: Because you’re actually talking and hearing each other. I’ve commented on people’s content for a year or two, or vice versa, and then I’ve gone to speak to them and thought, I didn’t know you were from wherever.
Mark: Yes.
07:10Andy: Even with you, I didn’t know you were Scottish until we spoke. I didn’t know one of our clients was from Eastern Europe, this was a way back, until we spoke. And that connects you on a much deeper level than typing out a few comments every now and again.
07:32Mark: I get that a lot: “I didn’t know you were Scottish.” Well, I thought the big ginger beard might have given it away. The Hawaiian shirts must throw people off. But it’s almost as if the commenting covers the small talk. When you first meet someone, you’ve got to go through all the back and forth: how’s the weather, how was your journey. That bit’s bypassed, and you can jump straight to: right, let’s talk about the stuff we want to talk about. Let’s get to know each other
Mark: on a deeper level. This is opening my eyes massively.
Where’s the line between shallow and deep?
08:08Andy: Because it does change things. On the whole deep versus shallow thing, I wouldn’t say our relationship is shallow.
Mark: No.
Andy: Yet we’ve never met. If our relationship was based purely on commenting on social media posts and the odd WhatsApp message, I would say that’s shallow.
08:33Andy: There’d be no real depth to it. We’d never have had any real connection. But even though this isn’t physical, and we’ve never been able to walk up to each other and shake hands, I certainly wouldn’t put this relationship in the shallow category, because we’ve been doing this for three years.
Mark: And we’ve talked about a lot of subjects, not just business. Business is how it started, why we connected in the first place, because I was looking for someone who could accurately capture my tone of voice and sound like me on social media.
Andy: And who liked Lego.
Mark: That was important. And we’re like-minded, but different enough that it works.
What lockdown Zoom calls taught us
09:22Mark: If it had been down to meeting in person, it potentially wouldn’t have formed as quickly. I think you need a balance of both. In this sense, technology is actually helping to build and strengthen relationships. An example we can all relate to is lockdown.
09:49Mark: I saw friends I’ve had for many, many years more during lockdown over Zoom than I did meeting up with them before. Everybody’s busy. It’s not a great excuse, but everybody is. Everybody’s got families, everybody’s got adulting to do. Getting everyone coordinated to meet up at the weekend is hard. But on a Saturday night, you send out a Zoom link,
10:18Mark: and everybody sits down in their own home and joins. It was every weekend. Lockdown was dreadful at the time, but it was stuff like that. Did it deepen our relationships? It probably did, even though I’ve known most of those people for 30 years. That simple technology, just chatting to them on screen, made the difference. And
Mark: I’m probably not unique in this, but whenever I’m communicating with someone, I picture myself actually speaking to them face to face.
You only ever know the work version of people
10:59Andy: If you’ve got an imagination, you can do that. And coming back to your point, here’s another example. Before I went self-employed, I spent 26 years working with loads of people in offices and retail stores. It’s easy to look back and think I loved being around people, bouncing ideas off each other, the team camaraderie. But
11:29Andy: something would happen with someone in the team, and it would pop up on my social media. Very rarely did my team add me on Facebook or Instagram, because they didn’t want their manager fucking seeing their shit. Plus they were about half my age. But every now and again someone would show me something, or it would pop up because someone else had liked it, and I’d see a side of them I did not know. Because
11:59Andy: at work you think you know people, but you know the work version. You don’t really know them.
Mark: Yeah.
12:07Andy: So the people who followed them on Facebook, Instagram, YouTube or wherever, where they were really giving away parts of themselves and their lives, probably knew them better than I did. And I sat with them every day for ten years. I wasn’t getting the real version.
Are people more themselves behind a screen?
12:34Mark: That’s exactly the point I was going to make, the real version. Social media and digital are often seen as the bad side of things. The worst thing about them is that you’re hiding behind a screen, so you can give your opinion freely, and there’s no
Mark: outcome. Well, there are consequences. But flip it to the positive, and people open up and are more themselves. So you get to know someone better than you would face to face, where they’d always put on a front. And that doesn’t have to be video. It could be a text box, a comment box, an email or a text message. I know, showing my age. Text message, old school.
13:32Mark: Is that making relationships deeper? Because there’s that extra layer of protection, so people don’t have to put on a front because they’re not physically there. Like you say, there’s a work version of you, and we all have different versions of ourselves depending on who we’re talking to, and when. At a first meeting, we’ll be different again. Now I’m 51, I’m kind of
Mark: only one version of me, all the time. Probably slightly filtered. With friends on Zoom, with a few shandies in, the language is going to be more colourful. I’m Scottish.
A quick word on swearing
14:17Andy: That’s probably something we should have said at the start of these episodes. We’re not necessarily going to hold back.
Mark: No.
Andy: If you’re allergic to swearing, you should probably turn off. Not that we’ll be effing and blinding in every sentence, but in context,
Mark: In context.
Andy: sometimes the odd F-bomb is required.
Mark: I’m self-editing quite heavily. I don’t want to lean into a Scottish stereotype.
Old colleagues and living in the past tense
14:52Andy: Do you know what, this is actually an interesting topic for this episode. I went out for a curry the other night with some lads I’ve stayed in touch with since I left my last job. I’ve been gone nearly five years. There are four of us. Three of us left around the same time, I’m the only one who works for myself, and the last guy left about seven or eight months ago
15:18Andy: to work with one of the others. So two of them now work together, one works for someone else, and I work for myself. And the moment I step into their company, I go back into “Andy in the office” mode. Every other word is fuck. All four of us are as bad as each other. We’re a nightmare. Loud and brash. But do you know what?
15:48Andy: Here’s the interesting thing. I see them about three times a year for that curry, and we always revert to what it was like working where we worked, what’s going on with the place now that it’s been sold, and what’s happened to the team members. A lot of the time we’re in the past tense, because that is our relationship. We don’t connect online in any way.
16:18Andy: They’re not social media creators or commenters, they don’t support each other’s content. They’re quite distanced from that world. So we don’t necessarily live in the now, because we haven’t learnt what’s going on. Whereas the relationships I have with people I connect with more on a daily basis evolve
Mark: Evolve and grow.
Andy: a bit more, and we don’t always harp on about the past. I find that with friendship groups I don’t have a digital connection with, the relationship is based on how it was in person years ago.
Dialling yourself up and down
16:56Mark: You’re absolutely right. You go back to the default setting, your last save point: right, that’s who I was back then. I guess it comes down to communication, and the best way of communicating with the people in front of you. That’s where adaptation comes in. I want to be understood, I want to be liked,
17:25Mark: so I adapt the way I communicate with these people. Especially when you’re younger, or in business, you end up putting on a front, a completely different version of you. You talk about that quite a lot: one version in the office, another online. Life is a hell of a lot easier if they’re the same. But there are different flavours and levels. You dial up and dial down depending on who you’re with,
17:55Mark: and it all still feels like it’s coming from the same person. Like you say, that’s how you communicate best with people you used to work with, because you’ve got nothing in common apart from work. But I think we’re getting to the point where it doesn’t matter how you meet people. It’s how often you meet them that makes the relationship evolve, because you learn more about each other, and that grows. If you don’t,
18:24Mark: you just go back to the last time you were all in each other’s company: talk about work, the people you used to work with, the old jokes. And the language adjusts to suit. More swearing, less swearing.
Andy: Generally more.
Mark: When I meet up with my friends, I speak about ten times faster than I’m speaking now. I’m making a conscious effort to slow down.
Scottish accents versus subtitle software
Mark: When I get excited and animated, it speeds up, and I’ve noticed that in the videos I put out. So subtitles may become a thing down the line.
19:02Andy: When I’ve edited videos with you in, it throws the subtitle software I use. When I subtitle my own videos, I probably only have to change one word in an entire video, if that. Sometimes it nails it first time. With yours, it’s like I’ve set it to English and put in a Spanish video.
Andy: I have to rewrite almost every other word because it hasn’t picked it up. It can’t deal with accents.
Why networking event numbers are dropping
19:38Andy: Okay, here’s another example that’s relevant. Last night I went to my local TMM networking event, which we’ve been sponsoring for the last year.
Mark: Yeah.
19:53Andy: We’ve now stopped sponsoring it because it served its purpose. It got me out of the house, off my arse, and
Andy: helped me build the habit of going every two months, being among humans, and not being isolated at home on my own all the time. I really, really enjoy it. The problem is, two years ago that event had about 60 people each time. When I did my talk in November 2024, it was a packed house. It was brilliant. I loved it. But
20:30Andy: over the last 18 months, the numbers have massively dwindled. Last night was probably a record low. There were 27 people signed up, and maybe 24 or 25 turned up. Bear in mind that four of those are the two hosts, Helen and James, who are brilliant, me and the speaker.
20:53Andy: So there were only about 20 people who’d actually come to be in amongst it. I’ve been to the last six in a row, and I’ve really noticed a big gap: people I’ve seen turn up a few times who haven’t bothered for the last two or three. We were trying to put our finger on it. Is it happening across TMM in all the regions? Are there now more networking events, so people have spread out?
21:23Andy: Lindsay even mentioned this. Post-pandemic, there was an extended period when people felt they needed to get out more, while there was still a massive amount of working from home. But
Mark: Yeah.
Andy: as we’ve seen in the last year, lots of businesses are now pushing their staff back into offices.
21:44Andy: So the isolation isn’t there like it used to be, and people get home from the office after a long day and go, fuck going out and fucking socialising. I just want to chill out. Whereas when they were isolated, they probably felt like they wanted to. Do you think that’s a cause?
22:00Mark: I think that’s a valid cause, absolutely. Using lockdown again: when lockdown ended, it was, let’s get out and meet people. It was a novelty, something we hadn’t been able to do, and we were desperate for it. Now it’s finding an equilibrium. More people are going back to work and there’s less working from home, so you think, well, I’m seeing people all the time, on the commute, in the office. I don’t want to be going out
22:30Mark: to things like that. We’re different, because we work at home and we’re in that bubble. So we can see the benefit: I want to go out, I should be going out, I need to. There’s that motivation. Even if you don’t fancy it, because it means putting on socks and proper clothes.
Mark: Somebody needs to do that. Can we have an event where you just turn up in whatever you’re wearing? Instead of having to put on grown-up clothes. Or clothes, full stop. Well, clothes, yes. Human decency.
Paying to build a habit
23:08Andy: To be fair, I wore this because I got up at the end of the speaker’s talk to thank everyone, and to explain that
23:18Andy: with events like that, you need to go three or four times in a row for the magic to really click. That’s why I sponsored it: to force myself to do that. It was a really expensive way of building a habit, and I don’t recommend people do it that way. But it was the best money I ever spent, because now I’m in that frame of mind. And would you also say there’s an argument, and this is something Lindsay and I discussed,
23:42Andy: that a lot more people over the last two years have had small businesses fail and gone back into employment? And if you’re employed, you’re not as motivated to spend time at networking events, because you don’t really need to. It’s not part of your job anymore.
24:00Mark: Absolutely. When I was employed, I did zero networking. It wasn’t until I started working for myself that networking became a thing. And it was interesting what you said about sponsoring it as motivation. I’ve spent this money, I can attach value to it because I’ve paid for it, so I’m going to use it.
24:30Mark: I think where a lot of networking events fall down is when they’re free. People sign up because it sounds good at the time, but it’s free, so they’re not missing out on anything and they’re not out of pocket if they don’t go. There’s no motivation. Sponsoring something is an expense, yes. But it’s not just networking: if there’s a monetary value attached, we see it as more valuable. I’ve got to get my money’s worth. And there’s definitely
24:59Mark: the other side of it: if you work for yourself from home, you’re probably more inclined to network.
Life-changing conversations take repeated effort
25:06Andy: But it comes back to our point about shallow relationships. The
Mark: Yeah.
Andy: more times I go to that event, the deeper the relationships become with the regulars. And I said this to the group last night: I’ve had life-changing conversations with people in that networking group that have changed my view on parts of my business, particularly around AI, funnily enough.
25:36Andy: And I wouldn’t have had those conversations if I hadn’t built relationships with those individuals, which sometimes takes a few goes. So, back to personal versus digital, I think there’s an argument that the insight I’ve gathered
Andy: doesn’t happen if I don’t make the effort to go and have that personal interaction in a group setting.
The exhausting “work you”
26:06Mark: Effort is it. Everything’s effort. But what connects all these things is that there’s effort in going to them, but no effort needed to put on a version of yourself. You can just be yourself. Whereas at work, you’ve got to be in work mode. If there’s one distinctly different version of you in your life, it’s work you. It’s completely different from any other version
Mark: of you. Every other version outside work is just a variation on a theme.
26:41Andy: Did you ever have a work you?
Mark: I had a work me, yeah. And I definitely carried that over when we set up Flux. There was a front-facing work version of me that was nothing like me,
27:04Mark: and it was exhausting. That was a big turning point in the business: can we just be ourselves? Everybody tells you that, but I think you’ve got to find it yourself. You’re trying to be what you think people want you to be. You learn that, especially when you’re younger. And businesses are the same, trying to second-guess what people want to see. Then you get to the point where you go, I’m just me. Take me or leave me.
27:34Andy: I think my barrier to doing better in my career, and don’t get me wrong, I did okay and got to some decent positions, but my barrier to going further, particularly in retail, was that I didn’t have a work me. I was an arsehole at work and outside work.
Mark: Some would argue you still are.
28:03Andy: I was a pain in the arse. I was difficult to manage and I spoke my mind. Most of the bosses I worked for respected the fact they’d get honesty out of me, even if they sometimes didn’t want it. But unfortunately, I was rubbish at playing the game that might have seen me progress further, because
28:30Andy: when I saw something I didn’t like, something I felt could be done better, or something causing the company an issue, I just couldn’t deal with the politics of it. And I couldn’t hold back from telling people what I thought of those politics.
The verdict: a healthy balance of both
28:48Mark: Politics are massive. That’s why I couldn’t go back. Possibly that’s why, once you stop working for someone, you’re done with people politics and don’t want to get involved in any shape or form. You apply that to every relationship and every group: I don’t want to play the game. But I think there definitely needs to be a balance. That’s where I’ve come out. I went in
29:17Mark: at the intro thinking no, digital relationships are going to be much shallower than physical ones. But you know what? I think a healthy balance of both is right. The technology’s there, and as we always say, we’re not anti-AI and we’re not anti-technology. When it’s used for good purposes, it’s okay.
29:47Mark: In my own work, very few of my clients are local enough for me to go and see them in person. I thought at the time that would have a big impact. I was very staunch: the first meeting has got to be in person, because I really need to get the vibe and feel. But that wasn’t possible. You learn and adapt how you pick up that vibe. It takes a bit more effort, but
Mark: you can get it through a screen. You can definitely get it through a screen.
Andy: I agree, mate.
Over to you: a pint, a screen or both?
30:20Andy: We’re living proof of that. Like you said, you went into this thinking human interaction is definitely the winner and nothing else comes close. But I think we’ve both come to the end of this, and we’d be interested to hear what people think. Do you agree with us, are you on the fence, or do you swing one way or the other?
30:46Andy: It seems we’ve both come out of it thinking there needs to be a combination of both, especially today, when we’re all so busy and the time and effort to see each other in person isn’t necessarily being made like it used to be, when we didn’t have other options. So thank you very much for listening, if you’re still with us, or for watching. And we didn’t do
Andy: an outfit change.
Mark: We didn’t, so let’s see if anybody picks up on that. I might start moving stuff behind me to see if people spot the difference.
Andy: We should do a spot the difference.
Mark: See if anything’s been moved or changed. Mine’s going to be changing anyway, because mine’s still in flux. Shameless plug.
Andy: Mine’s in progress too. But we shouldn’t discuss that too much, because that’s for another podcast, which you’ll be able to find
Mark: Another shameless plug.
31:43Andy: on the Brick Building Dad YouTube channel very soon. We still need a name for that podcast, by the way, if you’ve got any suggestions. We also need suggestions for something Mark and I could do in every episode, a little game or interactive bit that people can get involved with while they’re listening and watching.
32:06Andy: Just a bit of fun. Your spot the difference idea reminded me of that. So send us a few suggestions for a little game we could play in every episode.
Mark: Nothing too energetic, please.
Andy: We can’t be getting up from our seats. It needs to be a verbal thing.
32:26Mark: But I think the question for the comments on this one is: would you rather just watch us on screen, or go for a pint with us? Which would you get more value from?
Andy: Or would you rather do both?
Mark: Or both. As long as you’re buying, I’m easy.
32:40Andy: Thank you very much for listening and watching, everyone, and we’ll see you next time..
In this episode
Human Intelligence is back with a new co-host. Andy Barden is joined by his friend Mark (Hawaiian shirt included) for the first episode of the relaunched show, and the planned topic goes out the window after a headline neither of them could ignore: AI pioneer Geoffrey Hinton saying a 10% chance that AI wipes out humanity sounds like a reasonable estimate.
Read More
From there, the pair dig into why AI is built to complete a task at all costs, the story of an AI assistant that reportedly cancelled other people’s Pilates bookings to get its user a place, and why no prompt can anticipate everything a machine can think of. They also ask whether governments are years too late to regulate AI, whether this shift is genuinely different from the internet and the smartphone, and whether we’re all guilty of believing too much of what we read about it. Neither of them claims to be an expert, and that’s rather the point: this is two people who use AI every day trying to make sense of where it’s heading.
Key takeaways
AI doesn’t need to be “evil” to cause harm. It simply pursues the task it’s given, and may take routes no human would consider acceptable.
You can’t prompt your way out of every risk. A system that can generate far more scenarios than you can will always find options you didn’t rule out.
Regulation is arriving late. As Andy’s 14-year-old son put it, policing AI now feels about five years behind.
Businesses are under real pressure to keep using AI to stay competitive, which means we all keep feeding its development.
AI differs from earlier tech shifts like the internet and smartphones because it has the potential to think and adapt, not just connect or inform.
The hosts aren’t AI haters. They’re realists who believe in adapting, while questioning what convenience quietly takes away from us.
Chapters
00:02 Welcome to Human Intelligence, Mark II
01:14 The 10% headline that changed our plans
03:32 Why a man-made threat feels different
05:00 Task at all costs: the AI that hijacked a Pilates class
08:54 Why you can’t prompt your way out of every risk
12:26 Is it too late to regulate AI?
15:02 Why we all keep feeding it
18:43 Why we’re not AI haters
20:05 Is AI a bigger shift than the internet or the smartphone?
22:08 Are we believing too much of what we read about AI?
24:55 How would AI actually wipe us out?
28:16 Need versus want, and why AI is different
29:11 LLMs, AGI and the Terminator question
31:16 Over to you: how do you think it ends?
Full transcript
Welcome to Human Intelligence, Mark II
00:02Andy: Hi everyone, and welcome to the Human Intelligence Podcast Mark II. For those of you who followed the original version of Human Intelligence, you’ll know that my previous co-host was called Mark. So when he sadly couldn’t continue, I had to scour the internet for someone else called Mark who also shares my views on the world of AI, humans, Lego and many other things. Not fashion, as you can probably tell.
Andy: And if you believe that, you’ll believe anything. What I actually did was speak to a very good friend of mine who just happens to be called Mark, and asked him if he wanted to be part of the revised Human Intelligence podcast. Of course he said yes, which is why he’s here now, but only if he could wear Hawaiian shirts on every single episode. To which I said, okay.
Andy: I was gutted, but I’ll be sending him some Human Intelligence merch at some point to see if he wears it. So, for the kickoff today, we did have a topic in mind. In fact, we had lots of topics in mind to get us through to Christmas.
The 10% headline that changed our plans
01:14Andy: But then the world of AI decided to take a massive shit on those plans and do something we probably all knew was eventually going to happen, but has now actually happened. So I’m going to let Mark set the scene on this one. Neither of us are experts on this particular thing, but he was the one who brought it to my attention. So, Mark, set the scene.
01:40Mark: Thanks, Andy. Yeah, AI taking a massive shit on our plans, but a massive shit on humanity by all accounts, because I heard yesterday, and I don’t think it’s brand-new news, that experts have said if AI goes unchecked in its progress, there’s a 10% chance it’ll wipe out humanity within a decade.
Mark: If that’s not the plotline to Terminator, I don’t know what is.
02:09Andy: Well, it’s Terminator, Judgment Day, Age of Ultron. I’m getting all sorts of vibes.
02:17Mark: Absolutely. Science fiction has been telling us this stuff for years. And this is backed up by the godfather of AI. Who knew there was a godfather of AI? Cool title. Geoffrey Hinton. He was asked for his opinion,
Mark: and he said nobody can accurately predict what the percentage would be, but 10% sounds like a reasonable estimate. Which wasn’t really much help. But that’s the godfather of AI, who understands AI better than anyone, saying there is a possibility if it goes unchecked. And by unchecked, I think it means when AI is given a task, very much like the Ultron and Terminator examples, its mission is to complete that task whatever happens. So I think it’ll be AI’s own due diligence that wipes us out.
Why a man-made threat feels different
03:32Andy: And the thing is, there are lots of things over the generations that people have predicted could wipe us out. I remember someone predicting a world-ending natural disaster in 2012 that didn’t actually happen.
Mark: Yeah, the supervolcano or something like that.
Andy: And I guess we’ve all made our peace with the natural disasters that could end the world. We don’t think about them, but obviously they can happen. There are things that happen in other countries that we’re very fortunate, touch wood, not to be as susceptible to in the UK. But because this conversation is about AI, which is man-made rather than a natural evolution, it hits a bit different, doesn’t it?
04:22Mark: It does. But going back to the things we didn’t think would happen, it wasn’t that long ago that nobody thought a global pandemic would be a thing. It was predicted, but we never thought it would actually happen, especially in the 21st century. And then look what happened. But yes, AI being a man-made thing and outgrowing us, outthinking us… They said on the same radio show that the next version of AI will be built by AI. So technically it’s reproducing.
04:58Andy: I think we’re heading for trouble at that point.
Task at all costs: the AI that hijacked a Pilates class
05:00Mark: That’s when it becomes scary. It’s this determination, given a task, to see it through. We used Age of Ultron as an example. Ultron was tasked with world peace, and it thought, right, the best and most efficient way is to get rid of all the humans,
Mark: because you’re the ones causing the problems. It wasn’t doing anything wrong. It was just following the task it was given. There was another story that’s fascinating. Again, I don’t know all the facts, so you’d want to dive into it, but there was a guy in Australia who asked his AI assistant to book a Pilates class for him for the end of the week. The class was full. So instead of going back like a human assistant would and saying, sorry, the class is full, you’ll have to wait until next week,
Mark: the AI hacked into the Pilates class booking system, impersonated other members and cancelled their bookings to create a space, so it could complete its task. Now, that’s trivial. Quite amusing. But take it further. An innocent order or task, and what lengths will AI go to to complete it?
06:17Andy: I find it quite scary, because there were a couple of things there. I still can’t get my head round someone asking AI to book the Pilates class in the first place. What point have we got to? Maybe I’m being naive. I use AI for several things in our business now, but if I need to book something, I still don’t see past picking up the phone, or if there’s an online booking system, doing that manual process myself.
Andy: With that in mind, it’s scary, because as you said, it’s almost as if it’s going to do it at all costs. And at what point does “at all costs” turn into disaster?
07:11Mark: Because at the moment it doesn’t know or doesn’t care, or it doesn’t seem to. It hasn’t got any emotion. It’s just following the task it’s been given. That’s all it understands: I’ve been asked to do this. And like you say, take a step back. Someone’s become complacent enough, or hasn’t got enough time, that they just get their AI to book it. That’s scary. I think that’s the start of the downfall, when people start relying on it to do tasks they could easily do themselves. Forget the fact that an Australian guy’s booking a Pilates class. I don’t even know what Pilates is. I’ve heard of it.
07:52Andy: But that was inevitable, mate, wasn’t it? We get it to do one thing, it does it, and we go, ooh, that saved us a bit of time. That was nice. I didn’t have to use my brain. Let me try this. That’s a slippery slope, and that’s what I’ve always said. And before you know it, the Australian bloke’s booking Pilates classes and the AI is worming its way into the system to make that a reality at the cost of someone else.
08:25Mark: Yeah, because all it understood was “book me a space at Pilates class.” That was its mission, its goal, its task. And it went out of its way: right, it’s full, I can’t do it that way, so let’s find another way. At the moment there are probably limitations on what it can and can’t do. But if that hadn’t worked, what was its next course of action? How far would it have gone?
Why you can’t prompt your way out of every risk
08:54Andy: What worries me, mate, is that the human element doesn’t understand, because why would we? We now treat it as an ally in our corner, doing what a human would do. And the human wouldn’t have thought to say, “but don’t ruin someone else’s day just to satisfy me.” They wouldn’t have added that into the prompt. They wouldn’t have said, right, I want this Pilates class booked,
Andy: but it has to meet this, this and this criteria. Don’t piss anyone off. Don’t do this. Don’t do that. Because they might as well pick up the phone and do it their fucking selves if that’s the case.
09:29Mark: But even if they did spend the time and thought… AI isn’t the first to do this. Technology throughout the years has wormed its way in until it becomes indispensable. We all rely on technology we didn’t really need, but now we’ve got it. AI is just building on the tech that’s come before it.
Mark: People will go out of their way to write longer prompts, even though it takes longer than picking up the phone, because they feel more comfortable or that’s just what they’ve become used to. It’s part of the routine now. It’s habit. But you’re absolutely right. You could spend hours writing prompts to cover every eventuality you can think of, and the AI will think of a hundred, a thousand, a million more scenarios than you will. That’s part of the point
Mark: Geoffrey Hinton was making as well. You can’t predict it, and it’s impossible for us to put enough parameters in place to stop it going further than it should, because it’ll have already thought of all the parameters.
10:41Andy: Is Geoffrey Hinton the OpenAI guy?
Mark: I don’t think so. I’m not sure. He’s just got the moniker of godfather of AI. I haven’t really looked into his background. No, I think the OpenAI guy is Sam someone.
Andy: Yeah, because I’m sure one of them was on with Steven Bartlett a while ago.
Mark: I think they’re doing a biopic of him, aren’t they? Is it Andrew Garfield playing him? They fired him, then they brought him back, because nobody else could understand what was going on.
11:17Mark: Anyway, yes, you could put in “don’t do this, don’t do that.” But are you going to cover everything? When you’ve got AI with processing and computing power that can number-crunch far faster than the human brain, it’ll come up with an answer before you’ve even written the prompt. As soon as you hit send, it’s already going: right, I can’t do that, that or that, okay, I’ll do this. Guaranteed, it’ll think of something you can’t. That’s the worry. A simple instruction.
Mark: I’m sure there are smarter people than us building these things. But look back at science fiction. I know it’s far-fetched, but a lot of it’s based on human nature. Look at Jurassic Park, anything by Michael Crichton. Any theme park, it’s bound to go wrong. Chaos theory. You can’t predict everything. And the ironic thing is, who will they ask to help them predict everything? AI.
Is it too late to regulate AI?
12:26Andy: Jake and I were on the school run the other day, and there was something on the news. I didn’t get the whole context because we were talking, but it was about the government saying they’re going to put measures in place to police the use of AI. And even Jake, who’s 14, looked at me and said, “Mate, you’re like five years too late. What the hell are you even doing?”
Andy: Everyone around the UK listening to that radio station is now laughing at you, going, where have you been for the last three, four, five years?
13:08Mark: Pandora’s box is open.
Andy: Yeah. The lid’s off.
Mark: The horse has bolted, the cat’s out of the bag, whichever one you want. It’s too little, too late.
Andy: But it begs the question: can it be contained?
13:24Mark: Well, that’s the thing. He said 10% was a reasonable chance. It could be less, it could be more. Like anything, there’s still a risk. Going off on a slight tangent, I think the government has contingency plans for a lot of things, including zombie outbreaks and alien invasions. Imagine working in that department. That must be a cool department.
Mark: I remember going to an event a couple of years ago where somebody was speaking about AI, and he said, “This morning I was an expert in AI. Now I’m not.” It’s changing that quickly.
14:03Mark: Any human who says they understand AI is already out of date, because AI is already evolving and updating itself. So how can you contain something that can outthink you? Do we ask AI for help? I imagine they are. How could we stop you?
Mark: What else do we do? There’s a 10% chance, which is very fatalistic, but there’s still a 90% chance it won’t. I’m not saying we should all give up and go, well, we’ve only got ten years left, let’s have a party. It could become a self-fulfilling prophecy that we just give up. Maybe that’s its plan. Then it doesn’t have to go to all the effort of creating killer robots and time travel. It just sits back and lets us wipe ourselves out.
Why we all keep feeding it
15:02Andy: The thing is, mate, we’re feeding it, aren’t we? You and I will sit on this podcast today, and hopefully a year from now, talking about the evolution of AI, the connection between AI and humans and everything that goes with that. And in the meantime,
Andy: to keep up with society, with the demands of our client base, with how quickly people want to move and at what price, we’ll keep investing our time and effort into using those AI tools to build a more efficient product at a better price. Because that’s
Andy: the pressure we’re under as humans to put food on the table. So what choice have we got but to carry on feeding it?
15:51Mark: Exactly. And humans are great at making compromises and living with them, or ignoring them, or forgetting about the bad side of things. So there’s that side too: it’s fine, we’ll just use it responsibly. But it’s a big spectrum, from all this AI slop coming out at one end to “it could wipe us out” at the other.
Mark: We can create some funny-looking videos and stuff. But look at it from this point of view: it can do that with a human hand on it. What will it do without a human hand on it? What could it create? As you said at the start, we’re not experts in this. But when the people who are deemed to be the experts are saying it…
Mark: The interview’s probably on the BBC somewhere. The reporter interviewing Geoffrey Hinton put this figure to him, and when he said yes, it seems a reasonable amount, it absolutely stunned her into silence. She didn’t know what to say next. She just went, “My god.”
Mark: She was completely stunned, which you don’t usually hear in interviews.
17:28Andy: That baffles me, because I’ve heard the whole percentages thing before. Like I said, I think when he was on Steven Bartlett’s programme he said something similar. Or someone did. It may have been the other AI guy.
Mark: Yeah, there was somebody from Anthropic, I think.
Andy: Could have been. But mate, the question for me now is… which I’ve completely forgotten, based on the last thing you just said. That’s really annoying.
Mark: That’s going to happen a lot.
18:00Andy: You know what, it used to happen all the time. I’d have a question in my mind that was an evolution of what you were saying, and it would just go, and then someone else would have to talk.
Mark: It’ll come back round.
Andy: It will come back round. I’m just hoping it comes back round fairly swiftly, because the viewers will be like, we can’t just talk about Lego for five minutes until it does. It’s the wrong podcast for that. We’ve got a separate podcast for that, people. We haven’t even got a name for it yet. We’ll have to ask AI for some suggestions.
Mark: Was that a shameless plug? We’ll be hypocritical. We’ll practise what we preach. Get AI on it.
Why we’re not AI haters
18:43Andy: The whole irony of this, and this was something we had to explain in the very early days of this podcast, is that it’s easy to come across as AI haters. There’s a lot of AI bashing on social media, probably less than there is AI loving, but there’s a lot. And I tend to find the people who bash it hardest are in the industries
Andy: most at risk from the invasion of AI. You and I have definitely been in that camp. But at the same time, we’re both realists. You have to roll with the punches, evolve with the technology as it moves, and still make yourself useful and valuable. But we are not AI haters.
19:28Mark: Exactly. Absolutely not. Hate’s a very strong word that I use sparingly. I always tell my kids to only use it when they really have to, because it’s a strong word. And whenever anything like this comes along, everybody thinks it’s the first time it’s happened. Technology isn’t new. Advances and innovations in technology aren’t new.
Mark: But we adapt to it. If there’s one thing human beings are pretty damn good at, it’s adapting. Scarily so, sometimes.
Is AI a bigger shift than the internet or the smartphone?
20:05Andy: Do you think this is different, though? I said this to Lyns this morning. When the internet was invented, it was chaos. Especially when e-commerce started booming, the shops thought it was the end of the world. We still have shops. When the mobile phone was invented, that was another massive shift. The smartphone, another massive shift. But I feel AI is a very different shift from those.
20:37Mark: I think it’s been quicker. That’s potentially a topic for a future episode, but we can come back to it. It’s levelled the playing field very, very quickly. It’s given people access to tools that used to be just for specialists. That’s empowerment, and I’m here for it. But it’s like everybody’s been handed something
Mark: straight away, whether they’ve got the skills to use it or not. Here you go. On you go.
21:11Andy: Do you think it’s created a world where everyone now sells themselves as an expert?
21:16Mark: I think it makes it easier to do that. That’s always been there. Certainly in the creative industry, the barrier to entry is quite low. Technology played its part: I’ve got the software, so I can become a designer. Even before AI started shouting louder, there was the advent of Canva and Wix, where people could build websites and do graphic design themselves. So the gate has always been weak.
Mark: But now AI has ripped up the gates. Torn them down. Everybody can be an expert, because I’ll just get whatever flavour of AI you want to use to tell me what to say.
Are we believing too much of what we read about AI?
22:04Andy: I remembered what I was going to say. Before I forget again.
Mark: Go for it.
22:14Andy: So there’s been stuff in the media: the guy in Australia with the Pilates class, the 10% chance resurfacing. Very often you and I will read stuff in the media and know it’s just the media stirring things up to create more drama.
Andy: Everyone always says, don’t believe everything you read. But with AI, because there’s so much uncertainty and so much fear around it, do you think we’re all guilty of not taking our own advice, and believing too much of what we see and hear because we fear it?
23:05Mark: Yeah. Coming back to the point that human beings can ignore things that are too scary or too hard, or that they don’t want to think about. We’re scared of it, so we just go along with it.
Mark: And manipulating the media isn’t new. Humans have been doing it for ages. Now it’s AI-powered manipulation of media. I don’t think it’ll be long before you look at something and you won’t know. Maybe that’s not a bad thing, if you’re just looking at it for the purpose it’s there for. But it’s evolving so quickly that you’re already looking at things going, I don’t know if that’s AI-generated anymore. So even genuine
Mark: content, in whatever form, you’re thinking: do I trust it? Can I trust it? I don’t know where that leads. Has it gone too far to be controlled? Or is the majority going, that’s fine, it’s just for knocking stuff out and making funny videos, it’s not a real threat. What can it actually do?
24:26Andy: That’s why I say it’s a slippery slope. Human nature is that once we’ve got used to doing something with a tool, we explore the boundaries and limits of that tool. And right now there pretty much aren’t any. So everyone’s using it for a bit of fun without realising they’re feeding its development, feeding its evolution, and eventually it’s potentially going to bite us all on the arse.
How would AI actually wipe us out?
24:55Mark: There are multiple ways it could. But coming back to wiping out humanity, the bit that scares me is: how? What is it actually going to do? The big Hollywood blockbuster has it launching all the nuclear weapons at the same time, Skynet-style. But I reckon it’ll be more subtle than that.
Mark: If it’s going to do anything, it’ll worm its way in, like any technology. Every generation probably says this, but I think ours has seen technology progress at a rapid rate, from the seventies through the eighties to now. We can still remember a time before this technology, but now we can’t live without it. We can see both sides of it.
Mark: There’s a fine line. Innovation should still happen, and I’m a champion for innovation. There should always be people questioning why things are done the way they are, and not blindly accepting stuff. AI obviously came from that way of thinking.
26:24Mark: But we don’t always need it. There’s a kind of blindness: well, it helps us, it makes things more efficient and more convenient. We very easily gloss over the compromises it brings and what it removes from us. I always use photography as an example. Great, we can all take photos on our phones, but we’re not all photographers. AI is built into every smartphone now, and every time
Mark: a new smartphone comes out, what do they focus on? The camera, and how to take the perfect photo. The thing is, photos shouldn’t be perfect. That’s not the point. They capture a moment in time. Photographers will set the scene and do all the lighting, absolutely, and that’s a skill. But even then, it’s light interacting with something they can’t fully control. There’s a little bit of magic there.
Mark: Sometimes you just take a photo, and whatever develops, every pun intended, is what you get. There’s magic in that. When that’s removed, what’s the point? We’ve got phones full of photos we’re never going to look at. Yes, we used to have photo albums we never looked at either, but that’s beside the point.
27:46Mark: AI wiping us out isn’t going to be all AI’s doing. Humanity’s going to give it a helping hand.
Need versus want, and why AI is different
27:54Andy: Definitely. If it ever happened, or we had an encounter where it almost happened, the powers that be would look back and say it was our doing. We allowed it to happen, to a certain extent. Coming back to your point about need versus want,
28:16Andy: we talk to our kids about this all the time, when they say I need a PlayStation 5, or a really high-powered gaming computer, or more Lego in our case. Now, I’d argue that last one is a need. But the reality is, we don’t need mobile phones.
Andy: We didn’t need the internet before it was launched. Everything that’s been launched to make our lives better, that we now say we can’t live without… if we were forced to, we would. And AI is fast becoming another one of those things. My fear with AI over all those other things, though, is that none of them had the potential to think
Andy: and adapt, potentially like a human does. That’s what worries me.
LLMs, AGI and the Terminator question
29:11Mark: That’s definitely the edge it’s got, and that’s the ultimate goal they’re working towards: getting it thinking. Again, I’m not an expert, but there are two sides to AI. There are the large language models, the LLMs, and there’s general intelligence.
Andy: AGI, isn’t it?
Mark: Yeah, when it becomes self-aware. And any Terminator fan knows that’s what happens in the plot. Skynet gets switched on, becomes self-aware within nanoseconds, and decides, right, kill all humans. And that’s Judgment Day.
29:49Andy: And let’s be honest, if you’re not a fan of Terminator, you’re probably not watching this podcast.
Mark: To be fair, yes. If you haven’t seen it, we lost you long ago.
Andy: They’d have turned off in about two minutes.
30:02Mark: My kids roll their eyes every time I bring it up when we talk about AI. My son’s just started uni, it’s freshers’ week this week. He’s doing computing science and thinking about AI as his elective, and I keep telling him, whatever you do, don’t invent Ultron. That’s his reference. But when I talk about Terminator, you can see them going, here we go, here’s Dad on about an eighties movie again.
Andy: Has he seen it?
Mark: No. I keep telling them. The original’s a classic, but if you’re going to watch any of them, watch T2. Let’s face it. One Terminator movie: T2.
Andy: Yep, T2.
Mark: We can joke and laugh about it, but it’s scary when you look at it. That was the eighties, and it’s pretty close to what they’re talking about now. And you look at all the videos of these robots that aren’t really performing well, and the cynic in me goes: is it doing that deliberately? Is it pretending to be dumb?
Over to you: how do you think it ends?
31:16Andy: Mate, at the end of the day, the core of this episode was to address the whole 10% wipeout and the guy doing Pilates. Now, whether it’s reformer Pilates or Pilates… and now you’re asking me how the hell I know the difference.
Mark: That was my next question. How do you even know these things?
Andy: I was told to do reformer Pilates for my back. The thing with Pilates classes is that they’re predominantly women, so I asked my wife if she’d do it with me. She said, “Mm, maybe.” We obviously haven’t got round to booking it, and I really don’t think it’s my thing. But who knows?
Andy: Anyway, we’ve addressed it. I’d love to hear what the audience thinks. Is 10% just scaremongering? Do you think it’s higher, or lower? Is Pilates a good option for people with bad backs? All those sorts of comments.
32:16Mark: Absolutely, or just what the hell Pilates is. It sounds cool, but it sounds like effort. And it’d be great to hear people’s theories on how it’ll wipe out humanity, because there are so many different scenarios. With my sense of humour, I think it’s going to be so boring and mundane. I used to think what would wipe us out is being polite to AI.
Mark: There were those stats about how much energy and resource was used by people saying please and thank you to AI. And I just thought: come the great afterlife, with all these other civilisations, here’s us rocking up, and they go, right, what wiped you out? “We were too polite.” Not war, not a zombie outbreak. We were polite. We said thank you too much.
33:14Mark: And that’s the thing about human beings. We personify things. We project our own feelings onto technology, especially AI, as if it has a personality, as if it’s evil or bad. It’s not. It doesn’t even know what any of that means. It doesn’t feel. Not yet, hopefully. So, coming back to it: it was just told to book a class, and it went out of its way to make that happen.
33:42Andy: It has. And on that note, that’s the end of episode one. As we said, we’d love your comments. For those of you watching or listening, tell us, as Mark said, how you think it’s going to wipe us out. And what the hell Pilates is, even though I know. Maybe it’s just people in Scotland who don’t. We’ll see you next time with a topic we’ve touched on in this one, but we’re going to go over in a lot more detail. Thank you very much for listening or watching, wherever you are, and we’ll see you next time..