Interactions with AI.
The transcript.

Introduction

06:01 · Facilitator 1

You can see all of your advice flying around. Beautiful energy to kick us into our afternoon. And we're very, very pleased and super curious now to welcome Joakim to the stage. All I can say is that on your LinkedIn profile, you're called the AI and tech guy.

06:24 · JJ

Mm-mm.

06:24 · Facilitator 1

No, not anymore?

06:25 · JJ

No, not tech. Never, never, never tech.

06:28 · Facilitator 1

Oh, AI and internet guy.

06:30 · JJ

Sorry. Right. So what, what's the difference between internet and tech?

06:32 · Facilitator 1

Oh, gosh, Joakim. You're not quizzing me. Okay.

06:37 · JJ

Why should you get off so easy?

06:40 · Facilitator 1

Got off way too easy there.

06:42 · JJ

Yeah.

06:42 · Facilitator 1

But anyways, I met this gentleman, during a workshop, and I felt like all fired up. You're one of those people that really, like. Oh, I'm so curious. I need to know what makes you tick. That is Joakim. A round of applause for Joakim.

Bring the question home to yourself

07:00 · JJ

Thank you. Thank you, Chris. Actually, I just have to go back to this for a while because this is a great question to talk about. But for my session, you need to bring this home to yourself, right? So this isn't what anyone else should be doing. This is about what advice should you give to yourself tomorrow. So be in that space now for the next 50 minutes-ish. Okay? Cool. I'm also gonna see if I can just fix a little thing here. Let's bring that little guy in. And I'm gonna see if I can. Yeah, he's there. Cool. Okay.

Portable agents and a changing relationship with AI

07:44 · JJ

And speaking of little guys, how many of the, as we say in Swedish, small [unclear in source transcript] are here can you identify? Can we just shout out the names of any one of these?

07:56 · Unidentified speaker

Dot.

07:57 · JJ

Dot. I heard Dot. Yes, it's the one up to the right-hand corner. What is Dot?

08:02 · Audience member

Open, ChatGPT, open 24/7- Yeah.

08:06 · JJ

AI. Yes. New thing. Let's call it a portable agent, right? So it's one of those things that will live with you. So it's kind of the opposite to the Tamagotchi. Dot will be able to live in your little keychain, but instead of you catering to Dot to try and keep it alive, Dot will do its very best to keep you alive, right? So that's, that's the difference from the Tamagotchi. Do you recognize anyone else here?

08:27 · Audience member

Satya Nadella.

08:28 · JJ

Yep, Satya Nadella. He's probably also gonna try to keep you alive. What's next to Satya Nadella?

08:34 · Audience member

Muse.

08:35 · JJ

Muse.

08:36 · Audience member

Paperclip.

08:37 · JJ

There is Muse. Yes, Muse doesn't have anything to do with Satya Nadella. But the little figure to the right of Satya Nadella, if we start with the elbow, who is that? Clippy. Clippy, right. It's such a f- I, I don't know a- about Clippy. Who is that?

08:52 · Audience member

Microsoft Word.

08:53 · JJ

Yeah. What's that little fellow called? He's called Miko. For some reason, he's called Miko. Why didn't they bring Clippy back to life? I mean, all of us, I can see the age. I mean, all of us would have loved to have Clippy come back. And instead of just being dumb and say, "I can see you're writing an email. Do you want me to help with that?" Clippy would be able to do some amazing things to us. But no, Microsoft, in their infinite wisdom, decided to call their little portable agent for Miko instead. Breaks my heart. Someone said Muse. What is Muse?

09:30 · Audience member

Meta's, new assistant for you.

09:33 · JJ

Yeah. Like- Meta's portable agent. Yeah, correct. Finally, a reason to think that Meta is doing something good. Because that's kind of what it has done in the US. This is the fastest, app with 5 million downloads in the App Store in the US. Unfortunately, or thanks to EU, it's not here yet. Neither is Dot's because there are some privacy issues with those. But what, what Muse does, and what people have experienced with Muse, is that he lives in every device around you, and he is, she, it is, constantly tailoring for you and making your life easier. It's always. Muse has its own computer, for instance, and it's almost as a powerful computer as the one I have there, and it lives in the cloud. So Muse can work while you sleep or while you practice or, or whatever you do. And one of the more endearing stories was that one of those soccer fathers in the US who has a daughter who plays soccer, he was training his daughter's team, and they had an awful app that has been around for 20 years that they were supposed to use to handle, [unclear in source transcript] and practice times and, and so forth, right? And it's just horrible app.

10:41 · JJ

He sat down and talked to Muse for two hours, and now he has an app that not only works beautifully for him, but for the whole team and everyone around them. He just talked to Muse, and Muse made this problem go away. And the endearing part is, of course, that now that father thinks that, "Hey, now I'm gonna commercialize this, and I'm gonna sell this app to every soccer team around the world." The problem is every other soccer team is gonna do the same because that's how easy it is today to become a programmer. So the whole app industry is gonna, is gonna change fundamentally. And we have a few more. Do you know who that is? Oh. He's called Q, and all of those are called Q. So you can have Q in different iterations. That's an application called Manus.im. It was initially purchased or acquired by Meta, and then the Chinese government said, "no, we're not gonna have that." So they had to backpedal on that again. So now it's back to being a Chinese Singaporean app. But it's amazing in what it can do, and Q is really powerful as well.

11:41 · JJ

And then we have this guy down here, and I do think that's actually a guy based on its founder. It's GrokBot. It's Elon Musk's attempt to find his way into your pocket and into your lives. And GrokBot is actually one of those that you can start using today. So, doesn't really care about EU regulations. So you can, you can start using GrokBot today. The reason I show you this is that this is a whole new category that is coming and that is really kind of the foundation of how we will interact with AI going forward. You might love it, you might hate it, but you can't ignore it, so you need to start thinking about this. And this isn't a- about advice you should give to your organization. It's about what you should do yourselves. And I think you mentioned that we met two and a half years ago. Here is also like a portable agent, I think, someone you can keep in your pocket. I love how you're pointing to chaos there and said, "That's where we want to live," or, or rather at the edge of chaos, because that's where innovation and creativity happen, and so forth.

Karin, agency and the hacker mindset

12:45 · JJ

But the crew that is gathered here are seventeen executive assistants within Volvo Group. That's where we met, at this program called Next Step. And I was called into this program. It was a seven-month program for the best executive assistants inside Volvo Group. Those, you know, that are like the cornerstone of the whole organization. If they decided to stop working, Volvo Group would just crumble and fall. That's how important these people are. And in that program, I was called in or asked initially to come and do a forty-five-minute session on AI, and I said, "I refuse." Instead, we worked together for all of those seven months. We met once a week. We had something we called AI office hours, where they could check in with their problems, I could check in with some inspiration, and we could build together. And what happened during those seven months was that they went from feeling that AI is horrible and bad, and it's gonna take my job away from me, because that's what Gartner said.

13:40 · JJ

Gartner said that eighty-five percent of the repetitive tasks of an AI-- an executive assistant will be automated already by twenty twenty-five. So at this point in time, we were just one year away from them being kind of out of job. Instead, they started to love it, and so did I, because I. Here they are, by the way. That's the-- That I, I'm like the, the the, the wild card there, right? Everyone else, you see how homogenous they are, but they're also very, very diverse, of course, in the work they do, where they do the work, how they do the work. But one thing we found in common was that every one of them have some things in their work life that they hate. We called it pains, right? And we decided to kill those pains. Painkiller is one of the, you know, phrases we worked with. And my absolute hero is, you can see in the top right there, or top left from your side, Karin. And you can almost tell how, how cheeky she is, right? There's a big spark in her eye. And she decided early on to, fire Google and hire Copilot, for instance.

14:50 · JJ

That's one of her almost, you know, cognitive behavioral therapy triggers. If you want to change your behavior, you need one of those triggers. So she came to me, and she said, "I've decided to fire Google and hire Copilot," and that's what she did. Awesome. But she also showed me something else, and I'm gonna do something now that we probably shouldn't do. I'm gonna go live and see if I can show you what Karin showed me and the rest of the crew. So, if we go here, this is an attendance report in Teams, also known as total word salad that doesn't make sense to any human being. But her task was, after every meeting, and it was like three or four meetings a day, she had to work with this report and do this. So I'm gonna show you. She did that. She copied the name, and she went to Word, where she had a prepared template, and she pasted that name in, and then she went back here. And she copied the next one, and she pasted that one in. And let me know when this starts to feel slightly uncomfortable.

15:55 · JJ

Already? Yes. Okay. Now, imagine doing that three times a day for seven years and constantly asking for someone to, for the love of God, come and help me and fix this process and make it better. And I'm a geek. There may-- might be some geeks in this room who realizes that with the help of regular expressions and a Bash script or a Python script, you would fix this in three minutes. Do you think anyone at Volvo Group Digital & IT thought that this was important enough to fix for Karin? No. I mean, there are only twelve thousand people. How should they find the time to help her with this? It, it came to the extent that she actually called it out as an arbetsmiljöproblem, you know, when they do those Gemba walks, right? [A personal remark about a colleague’s distress has been omitted.] Like, this is not human work, and no one cared. Enter Copilot. So what happened here, I'm gonna take that away, so we don't have Word any longer than necessary. What she realized was that if she takes this and do one copy, so she took all of it, all of it, and then she went to her new friend, Copilot, and she said something to the extent of, "Create a, a bullet point list with just the names." You see very advanced prompting skills here, and then she pasted in the list, and I think you have already figured out what happened, right?

17:20 · JJ

She got this list. So going from multiple copy and pastes into one. Can you realize the agency that Karin feels when she realizes this? And this was one of her very first interactions with AI, and she instantly recognized that, "Here's something that will change my life. It will change how I work." And since we had those AI office hours, Karin shared this to the other sixteen executive assistants, and you know what happened? She became the hero, of course, right? She, she was like their new AI leader, right? Screw Volvo Group Digital & IT. Karin, how can you help me with the next thing? And since so much good happens when people start to interact with each other and have conversations, someone, of course, said, "Well, Karin, this is excellent, but the next step I have to do is that I, I have to, turn this into an, an, a send list in Outlook, right? Because the email should only go to the ones who were actually in the meeting." And Karin said, live in the meeting, "Well, let me try to figure that out." "That's great, perfect.

18:24 · JJ

But I also want a list with email addresses formatted, so I can just paste it into Outlook and send the email away." And I think you already figured out what happened, right? She got this perfect list ready to copy and paste. And the next iteration over the coming weeks was, of course, to tailor this into one prompt that did all of it, save the prompt and share it with everyone else, figure out a way to not have to copy and paste the text, but rather just slash and point to it on the disk where it already was. And of course, the third step and, and the ongoing process here once they got access to that was to create, an agent that did it for them. So they could just call up the agent, they could do something like this and send it away, and the agent did all of the work for them. So going from multiple steps to just one step, basically. You have to think about what this means for a human being who had never had the capability to solve these kind of problems before. She had lived a life where her interactions with digital was to push the buttons that someone else had created for her, follow the instructions that someone else had given her, and now suddenly she was the champion of her own destiny here.

19:41 · JJ

Does that make sense to you? Cool. So we survived the first live demo. This is my takeaway, how we will all be hackers, and this is the definition on Wikipedia of what hacking is. It's the process of exploring and manipulating. We can-- Let's stick with exploring computer systems. It's often associated with understanding how systems work, troubleshooting, finding creative solutions to complex problems. I agree one hundred percent, but then I stop agreeing because it says hackers are often skilled programmers or engineers. Hackers today is someone like Karin. Hackers today is someone like that soccer dad who created an application that solved his problem. And not only will all of us be hackers, we will also be leaders, and not just in the way, shape, and form that I told you about Karin here, where she could lead her other fellow executive assistants, but also in leading AI teams. I'll show you what that looks like, but first, we just have to make do with a misconception. This got a lot of weather, I think it's a year and a half ago, the MIT report that said ninety-five percent of generative AI pilots at companies are failing.

Learning, pilots and the five-part progression

20:55 · JJ

That's bull crap. It depends on how you define it, and even if you read that MIT report, that's not what it said. But media, of course, they want to-- want you to subscribe, they want you to, to. I, I think there's something up here. What? One hundred and twenty ad cookies, et cetera, right? So there is a big incentive today in scaremongering and talking badly about AI and chasing those headlines. Now, I'm not saying that AI isn't problematic. It is. That's for sure. But we need to compartmentalize. We need to figure out what actually matters to me and what can I bring into my circle of influence. And there is absolutely nothing that we did during those seven months with all the executive assistants that would classify as a fail. That doesn't mean that everything was a success, but everything was a learning, right? So either they succeeded or they learned, and that's the mindset that you need to bring home tomorrow. So why do we sometimes fail? Well, I'm s-- I'm sorry to bring it to you, but we are the problem.

21:59 · JJ

AI today is in such a good shape that we can actually have it do whatever we want if we only have the right intention, the right, capabilities, and the right vision of where we want to go. Toolset, skill set, and mindset. And I'm gonna focus on the mindset today because that's where it all happens. So when we talk about human AI and, and human-AI interaction and, and, you know, interacting with AI, we've already done away with discovering agency. That's the Karin story. I'm also gonna show you what happens when we go from com-- complex prompts to engaging conversations. A mediocre human with a great machine, I'm gonna explain what that is all about. Kasparov is here to join us. Mutual growth through honest feedback and, from co-intelligence, to do-- coexistence. Sorry. What happened there? Let me just. That was me. That wasn't the AI who effed that up. There we go. Coexistence. That's how easy it is to fix a problem. Now, I'm gonna explain the other four in more detail, but we don't have a lot of time today, so I'm also gonna have to tease you to go into the post-session material that we-- you will get your hands on, right? So everything we talked about here, including transcript from everything that happened in my session, we're actually gonna work with that. But I need to. Let's go here first, right?

From complex prompts to conversations

23:32 · JJ

So let's, let's see how we can go from a complex prompt to, to talking with AI. And we all know what complex prompts might look like. So I'm gonna show you this, right? This could be a complex prompt. I was actually the, the first runner-up in the Swedish Prompt-SM two and a half years ago, so I know about prompting. But I strive so hard to forget everything I know about prompting today because what I do instead is something to the extent of this. Hey, Petty. Nice to see you. Can you just explain to me what you're seeing on the screen right now?

24:17 · JJ

Well, that turned out well. There's something with audio here that has been haunting us the whole day. So, let's, let's forget about that. And instead, let's go That's a shame. That's a shame. We'll see if I can find a way. I'm gonna do a screencast for you guys tonight where you can see how I interact with my AI today. But anyway, that's, that's a general gist of it. All the AI tools that you have today, including Copilot if you're a Microsoft, environment, you can bring up in your phone and start having conversation with them. You can do triage of your email while taking a walk. You can even, in the car, if you have CarPlay enabled, you can talk to Copilot in the car and prepare for the meeting that you're already late to, and so forth. You know, the endless hours that you spend trying to find a parking lot around the Volvo ground-- Volvo complex, you can actually be useful there now with the help of Copilot. So start talking to the stuff. I'm not gonna dwell on that. Instead, I'm gonna tell you the story about Kasparov, because he was like the, the crown of creation. He was the jewel of mankind.

Kasparov: human, machine and process

25:25 · JJ

He was destined to be known as the most intelligent human being ever born. A great chess grandmaster. This image is from a game. I'm gonna see if I can just take that away. This image is from a game that he played in 1996. Do you know who he met in that game? Deep Blue. Deep Blue, correct. And Deep Blue was a new kind of chess computer. So we've had chess computers for the longest time that were able to kick the butt of every one of us in the room, but it couldn't defeat Kasparov because besides being an incredible machine of memory and pattern recognition, he also had creativity, and the machines didn't have that. But in this game, this is called his moment of despair. When he looks down at the chessboard, I think this was move eleven, and he realized that, "I'm screwed," right? "There's no way I can win this." But he also realized that the machine had made a move that no human being had ever made and probably would not ever make because it didn't make no sense. But he knew enough about chess and pattern recognition to realize that from here to the end of this game, I can only lose. That's his moment of despair.

26:36 · JJ

And based on that, he came up with, let's call it three laws. He wrote a book called, called, Deep Thinking, and I have extracted three laws from that. And the first one is that a machine will always beat the human being, and that's kind of the reason for despair, right? We-- Whatever, if we go narrow enough, like chess or like, image recognition in cancer diagnosis, we will always be beaten by the machine. His second law is where we get some courage back, right? Where he says that a machine and a human being working together will always beat the machine. That's not really true anymore. We human beings might get in the way from a better decision made by the machine. But the bigger problem with this is that we might chill a gorilla a little bit too much, right? We might say, "Well, I use ChatGPT," or, "We have Copilot in my organization, so we're done. Everything is fine." But his third law is where it all makes a difference, right? Where he says, "A mediocre human with a great machine will always beat a great human being with a mediocre machine." So he says that even if all of us are individual, when we boil it down to our core capabilities and skills, we are not so much alike that we can see ourselves at the-- as, as the one that gives leverage in a combination of human and machine.

27:49 · JJ

And then he says, "Add process to this." So what we need to take away, and what you need to take away going home tomorrow is that, not only do you need to figure out what is the best machine for the job to be done, but you also need to figure out, how do I do that? What is my process of interacting? It's a little bit like trying to speak French to someone in Spain, right? If you haven't figured out the, the common language and you haven't figured out how to interact with AI in a solid way, you are not in a very good space. The good thing is-- there is that the only and the best way to learn this is to actually do it. Aristotle said that what we have to learn to do, we learn by doing. It has never been as true before it is today, because once you start talking to your AI, it can guide you. It can help you. It can be that twenty-four-hour ever-present coach and tutor that you can lean on, and you should. Learning together, this is your new learning buddy. Now, the next thing is, is about feedback, and, and I have to, I have to start here. This is me, some forty years ago.

Feedback and the AI team

28:53 · JJ

I did seven years in the army, and the one key takeaway, and the one thing I miss the most from there is giving and receiving feedback. It's so incredibly powerful. Feedback is a gift. But in the civilian life, n- it doesn't really matter if it's public or private sector, human beings are so scared of feedback that they really try to, you know, stray away from it, right? They try to avoid it at any cost. But back there, we-- it was so instrumentalized. It was so structured. We were expected to, every day after every event, give feedback upwards, downwards, sideways, and accept it and take it back like the gift it is. I've missed it sadly ever since. So let me show you then what that can look like. So first of all, I'm gonna show you my team. Where is my team? Here is my team. I have, fifty-two different AI services in my team that I work with. I have orchestrated them together. There are multiple accounts. The total salary that I pay to my team every month is five thousand nine hundred and thirteen dollars, and that's probably a lot of money to a lot of us, right?

30:05 · JJ

But if we compare that to what a human team would cost, this is kind of the equivalent of what my AI team does for me. It's a whole lot more, right? And if I look at the ROI that I get from the output of my team, it's a lot. But we can, we can, we can turn this into money any which way we want, right? But the core thing here is that I get something more than just efficiency from my team, just as I would from a human team. This team here gives me a lot of feedback. And one way it does that is that I have something called a lot of, you know, attention watches. You probably have this, all of you, right? So this is something that, is handed to me basically every morning with an overview of what I need to prepare to go into this day today. And one morning-- I'm gonna show you what happened here. One morning, my team gave me a really bad prep. It was awful. There were so many holes in the prep that I got going into that day, and I kind of blew a fuse, right? I got mad. So I spoke to my AI and said, "What the F are you doing? What's this?" and what do you think happened then?

31:17 · JJ

The AI came back and called me out and said, "This isn't on me." You-- The AI said, "This isn't on me." It pointed to me, the human being, and said, "This is you being a D-level decision maker, because I have asked you for three straight days now how we should handle these gaps, and you haven't answered. So it's not my fault, it's you." Can you imagine getting that from your AI team? We're not used to that, and, and neither is AI. So a bit down that thread, it kind of realized that it's not supposed to be this rough to me. It's supposed to be much more servant. So it softened the whole story, and, then this happened. I-- So I said to my team that I love feedback. You are-- This is me, right? So I said, "First, I don't mind direct feedback, even if it's harsh. Calling you a D-level decision maker is like calling a spade a spade, and it's okay." This is me having a conversation here with my AI. And then I followed up with, "As long as we don't stop there, right? As long as we discuss how to get on with it, how to improve, how to find ways of working, it seems to be exactly what we're doing right here and now." And then I said, "But for now, put a pin in it because I'm busy for the rest of the day.

32:27 · JJ

Let's dive into this later tonight. High priority." This is a conversation that I remember so well from my years in the army, right? We had to park it for now and then get on with the conversation and try to learn from it. This is what my AI said back to me. "Absolutely. Then I'll keep the gloves off." It understands me so well. And then it said, "The useful standard is not, you know, yada, yada, yada." And then it said, "Tonight, I'd like to go deeper on three things." This is-- Again, remember now, this is my AI talking to me. It said, "What the evidence really say about you versus the system, which recurring failures are behavioral, structural, or incentive problems, and what is the smallest operating model that would kind of actually change the trajectory that we're on so we don't wake up every morning and clash about how bad it is or how bad I am?" I love this. To me, this is absolutely amazing. I grow every day by these kind of interactions with my AI. But how about you? How would you feel, Hanna, having that kind of interaction with your AI?

33:38 · Facilitator 2

Oh, that's such a good question. I would probably be a little bit, like, struck in the, in the beginning. But I agree with you. I love feedback, so I think. But I will be challenged. No doubt about it, I will be challenged. But I think I would lean into it after a while, and I think it's useful. Mm. But we're curious about, well, how would you feel about getting this kind of feedback from your AI team? What would be your immediate reaction to getting feedback like this and having this kind of feedback relationship with your AI team? Turn to a, a person on your table and share. What would be your reactions? Give it a few minutes.

34:29 · JJ

Yeah. And you can work on that.

[Table discussion. Incidental fragments are omitted; the plenary resumes below.]

36:54 · Unidentified speaker

Okay. All right. Lots of conversations here.

Audience discussion: feedback, trust and memory

37:24 · Facilitator 2

Let’s see. I love the energy. Is there anyone that would love to share? Okay. I love the energy. Lots of conversations here. Is there anyone that wants to share their thinking? Raise your hand. Come on. Oh, we have somebody. Okay.

37:48 · Audience member

[Participant names and affiliation omitted.] We were talking about this, and we were going back to JJ's first point why he loved feedback in the army, and we thought, "Why was army so good at giving feedback?" Because if, you know, things go wrong in the army, then it's a really shit situation. So it has to be very good, and you have to be very prepared. And the fact that we miss it in our organizations is so sad because at least for both of us, we felt like we don't get that feedback. But what if AI can be the bridge between the manager and the person or, or between two people and make the blow a little bit softer, but still make sure you get it? Wouldn't we make our outcomes so much better, like in the army, so we don't have to die for no reason?

38:40 · Facilitator 2

Oh, I love that perspective. Excellent. Thank you so much. Any other voices? Let's see. Someone else. Yes, we have one more.

38:51 · Audience member

We're talking about, with good feedback often can help if.

38:57 · Audience member

Yeah, with good, good feedback, it helps if there's a high degree of trust. And one of the-- In my use of AI currently, mainly Copilot, I don't have the trust because I feel like it's got no memory. Sometimes I think, "This is off." Like, hang on, we're just talking about that. You've forgotten it all. So I don't feel I could have a continuous relationship like you, you kind of shared there at the moment in my, my use.

39:21 · Facilitator 2

Okay. So there's the trust part here. Yeah. You want the continuous relationship like that. Well, yeah. You have to build it then. I, I need it. You need it. Yeah. Thank you. Okay, one more voice. Okay.

39:34 · Audience member

So a question, how did you get it to give you that direct feedback? Because we were talking about AI is always so nice and polite and so on. How did you get it to give you that direct feedback being a D-level decision maker?

39:49 · Facilitator 2

That's a good one, Joakim. Over to you.

Context, memory and feedback in practice

39:51 · JJ

Yeah, excellent. And, and the, the honest answer is that by interacting with AI, I asked it, "How can you be more direct?" And I-- We had a conversation about my level of feedback. And once it understands that, it can actually be really harsh back, right? So it, it's, it's all a matter of how you, how you tailor it. And, and to get back to the memory part here, I-- Let me just show you this. You mentioned Copilot, right? And I think one of the reasons why we often experience Copilot to be as useful as a chocolate teapot is because we haven't enabled the capabilities in Copilot, either because you haven't found them or because your IT department, in their infinite wisdom, have decided that we can't trust our colleagues with these capabilities. But once you enable stuff like, say, memories and chat history, it is no longer like that really stupid colleague of yours that comes in every morning like Groundhog Day, right? And, and not understanding anything on what happened yesterday. So with those two simple buttons, you can actually turn this into something that is much more, useful for you. So enable those.

40:56 · JJ

And then as we move forward into more and more advanced capabilities, so now, for instance, we do have, here we have something like Cowork that we could enable inside, Copilot. And what it can do once you enable that is that it can start writing files to your disk, right? And that's basically the same thing as creating its own memory, creating understanding of your work processes and all of that jazz, right? So that's a little bit of what you can do in Copilot. Don't give up on Copilot. But of course, there are so much-- so many better tools there. And I brought this up also to answer your question, because having it be harsh and, and direct is one thing, giving it enough understanding of what you're doing is another. And Copilot has two simple switches. In ChatGPT, in the, in the application that I run on my Mac, I can actually enable something called computer history. And by the means of the accessibility settings in, in the Mac that is used for people with, you know, accessibility issues, where they can't read and they can't hear and whatever it is, there is a whole bunch of capabilities in my Mac that makes it possible for Copilot or for ChatGPT in this case, to see all the messages that I send, see every activity I, I do in Chrome, where I browse around, to look at all the applications that I worked with here this, this afternoon or what I did this morning.

42:23 · JJ

And with that knowledge, two things can happen, right? One is that it can suggest that this looks like something you are-- you're doing all the time, creating an AI prototype. Let's create a skill to do that, so the next time you ask for the same thing, it's gonna be much more fluent. And this is-- I, I didn't prompt anything, right? I just gave it access to what I do, and it figures out that here's something that we could probably make in a better way or a more efficient way. But I can also interact with it. So if I go all the way up here, I can chat with my computer history. Who contacted me today that I need to follow up with? Well, that's a boring one. Let's do What are the main behaviorals that I should change to be a better manager over you, dear ChatGPT? So what it's gonna do now is look into my computer history, look into the files that I've saved, look into the memory that it has captured around me, and it's gonna give me an answer to that question. And every time I do this, every time I interact with AI like this, I learn, right?

43:27 · JJ

And I learn my AI about how I think that things should be done. So it's really truly an interaction. And I know that we shouldn't humanize AI. We should be careful. We should be well aware that it's not a human being. But in how it operates in this process, I can honestly say that it's one of the best human beings that I have, you know, interacted with because it's so good at doing stuff like this. "Your biggest improvement should be to delegate completion more clearly." We're back to that D-level decision-making thing again, "And to spend less time personally coordinating the work." I reviewed computer history from September twenty-fourth to 30th. There are interpretations of observed activity, and here we go, right? So here is my performance review or my one-on-one with AI on what behaviors it think that I should change to, to improve. And this can be really soft stuff as well. I have no problem letting AI into my personal life, which also means that it can give me advice on how I should better interact with people, how I should, you know, have conversations with my wife and, and basically everything.

Value beyond efficiency and the four jobs

44:38 · JJ

Are you with me? Yeah. Good. Okay. So in the material that you get tomorrow, you're also gonna see stuff like this, right? Because when we talk about AI today, we're so focused on efficiency, and I think that's a problem because the majority of the value that we can create live somewhere else. Specifically, I would say in, in context like this, it lives here in access and inclusion, in how it can help us human beings to be the best human being we can be. For instance, by taking away our issues with typing and, and, you know, writing stuff because we might have dyslexia, we might be anywhere on that. In general, this is just an amazing opportunity for us to be the best we can be. Another slide is this, where I do know that most of us work in organizations that are incredibly performance-driven, right? It's all about getting the job done. What we need to do, and as we did so well with the executive assistants on Volvo, was to give them an opportunity, recurring, put it in the schedule, to actually think about how do we do the work, how can we improve this?

45:48 · JJ

And I said that Karin did that task three times a day, and it took her about five minutes every time she did that. That's fifteen minutes a day. How many work days is that on a yearly basis? And I'm, I'm not gonna talk about compounding here because I, I heard this morning that we're not supposed to use that word. But fifteen minutes a day for a work year, how many days is that? A week. Your AI would answer in a blink, right? It would say seven and a half work days. That's seven and a half days that Karin could spend here and now to improve the next thing that makes her life better, or just have a conversation with AI to get feedback and grow herself, right? Or how can I help grow my team? So every minute that we save or get back from AI doing something for us, we should spend here. And if you're a manager, if you're a leader now, you need to figure out, when my staff, when my direct reports, when my team is so efficient in using AI that they're done with the job after four hours, what am I gonna do?

Human judgement and the four guiding principles

46:57 · JJ

Am I gonna send them home or am I just gonna pour more work on them? How do we handle that situation? So there's a lot of leadership questions in this. And to kind of start closing this off, what I'd really like you guys to do now is to take this on board and go home and start inviting AI into everything you do because that's the first drug, right? That's the first entry point to all of this. Be the human in the loop. We're not supposed to say that either, I guess. But if we take this into not having you as a, as a human being having to check off everything that AI do on the way out, but rather using your judgment to design the processes, to make sure that. You know, just like situational leadership, right? In the beginning, you need to be really controlling. You need to be on top of what your team delivers, and then over time you learn their capabilities, you learn what they're good at, you, you know, you spread the tasks, and over time you're able to trust them and let them work in peace.

47:58 · JJ

That's what human in the loop means to me. Treat AI like a human, but never forget it's not. And this is usually about, you know, it, it might sound like it has feelings, but it hasn't really. Yes, that's true. But it's also a matter of how you lead your AI team because when we lead human teams, we need to be a little bit soft in the, on the edges, right? If. Let's say for instance, I have a human team. I send them off, to do a task for me, and they come back three weeks later super happy about what they have achieved. And thirty seconds into when they present what they've done, I realize that they have gone the total wrong direction, right? This wasn't at all what I expected. We start to compromise because we don't want to be that douchebag. With AI, you can be that douchebag. You can be direct. You can be harsh and have it be harsh back to you. So we need to rewire ourselves, right? So that's that kind of the paradox as we start to look at AI as more and more human, as we start to interact with it in more and more human ways.

48:53 · JJ

Don't fall in the trap of having to be that good person all the time, right? At least once in a while you can be a douchebag. Yeah? And the f- the last one there, you know, Copilot, yes, that's an excellent example of how it was really bad just six months ago, how it's today just bad, but how it's definitely getting better, right? So you need a parking space. Everything that you try to do with your AI team that they fail on, put that on a parking space and revisit it a week later or a month later and see if the AI has gotten better or if you have gotten better, right? I wanna show you one more thing before I, before I end, and it's about. I've, I've already shown you how you kind of put yourself out there by giving AI access to everything you do, but we can actually take this one step further. Interactions can also expand beyond you. And let's again take a human analogy. You have a new coworker joining your team. Initially, that coworker knows nothing about you. You need months and years to have them understand what you think that good looks like, what's your kind of love language, and what's your communication style. And so it, it takes a long time.

Public context and the Lagunen demonstration

50:12 · JJ

The fact is that the same thing happens with AI, right? It, it's a process to have AI understand you, and even if we flip those switches in Copilot, it takes time. But what you can do is that you can shortcut this process, and I'm gonna show you what that can look like. So here is, here is my-- I call it a, a first party public context. So this is how I describe myself. This is, my family. This is what matters to me. This is how you should work with me in practice with my calendar link and you know how I love asynchronous meetings because I have weird working hours. I have-- This is my, my idea of learning, curiosity, and play and so forth. So this is an extensive profile that makes no proposal to be objective. This is how I describe myself, and this is how I would hope that through feedback and working together, a coworker would think of me in the future, right? Once we have gotten to know each other. What I've done here is that I have, of course, placed this online, and I have given you a prompt here. So for instance, before every first meeting I have with someone that I know works with Copilot or whatever they do to prepare the meetings, I say, "Copy this ph- this phrase into your AI before you start working with me." Because then their AI, not just mine, but their AI will have an idea of what I think that good looks like and how I think that we should work together.

51:42 · JJ

Does it make sense to you that you can, you can, you know, put yourself out there by doing this? And I love the deep interactions. I love being able to have a conversation with my AI or with your AI and, you know, back and forth and so forth. But there's also another aspect to putting you out there like this. And I actually built a demo tonight, last night where I play that I am a camping site called Lagunen, and this is their, CMS, the cus-- No. The-- What's the name? Where you handle your customers.

52:21 · Audience member

CRM.

52:22 · JJ

Right. Cool. So this is it, and here's an incoming question where it says, you know, [On-screen Swedish enquiry partly unclear in the automated transcript.] So that's an incoming question. And what I've done in the signature of all my emails is that I have said that you are allowed to read my public profile so they can send their AI to read this. And let's see the difference in the answers here, and hopefully this will work now because this is very much live as well. So here's what they would respond ordinary with just the incoming cold email, and this is what they might respond when they have access to my public profile and can learn more about me. Let's see. Here we go. Yeah, it has taken a lot of account to what I think is important. It has used my link to suggest an, an upcoming meeting where we discuss more about it. Let's take something more personal because that's more fun. "My name is Joakim Jardenberg, coming to you with all the family.

53:27 · JJ

We haven't decided on dates yet." Let's see what that looks like. But do you see the potential in this, how you can basically have better interactions with the whole world by putting yourself out there in an AI-formatted way? That's, that's a tough one, right? Oh, here we go. So that's-- Let's do this. Same leg as you can see it clearly. So that's the ordinary standard answer that I would get back. Cold email, cold answer. Here's the warm one. "Hey, JJ. Om du med hela gänget menar du och Kattis och Brian och Nelly och kanske Tina, så skulle vi börja med välja den här platsen. Eh, då skulle vi kunna laga middag tillsammans för den vet jag att du älskar det." Och sen så säger den, "Another setting would be to, since Brian and Nelly are newly married, maybe they get their own place to stay. They can't promise." And there's my old mother-in-law. Here's examples of what we could do. Here's the dog. If the dog joins, you need to make sure that that's a part of your booking. And here we can charge our electric car, and again, it has used my link too.

What will you change tomorrow?

54:32 · JJ

Do you see this? I know how uncomfortable we might feel about this, but think of it like taking control. Rather than just letting anyone out there fill the gap with information about you, by putting a profile like this out there, I take back control. Interactions comes in many colors, and I know that most of you fear some, some things here probably, but once you start doing this, the fear will go away, and, and you will love it just like I do. Now, here's a question, I think. So we have a few moments, right? What will you change tomorrow? And I've given some examples based on what I talked about, but you might just as well decide to do something else. Hannah?

55:17 · Facilitator 2

Yes, I will definitely be even more intentional with my AI. I need it to be more direct in its feedback. But we are of course curious, what would you do different tomorrow? So again, think about this. Something that you will do different. Share with the people around you in a moment or two. Put some words to this.

55:35 · Unidentified speaker

Mm-hmm.

55:36 · Audience member

Beautiful.

[Table discussion before the closing remarks.]

Closing

59:03 · Facilitator 1

Okay. Welcome back. These conversations need to continue. Sorry to interrupt you, but you can imagine how buzzy the room is gonna be now, right? Bing, bing, bing, bing, bing. You can see all the pens flying and all the conversations happening. You're going to continue those in a break in just a moment. But first, this man is not only inspiring, he's also quite simple to find online, right? Mm-hmm. Now that you can see his public profile. So if you wanted to work with him, there he is. A round of applause for Joakim, everyone.