There’s No Better Time to be a Change Agent Than Now, with Paul Acito, CEO at Lyftbridge Innovation

Transcript

Our transcripts are generated by AI. Please excuse any typos and if you have any specific questions please email info@digitalshelfinstitute.org.

Lauren Livak Gilbert (00:00):

Welcome to Unpacking the Digital Shelf, where industry leaders share insights, strategies, and stories to help brands win in the ever-changing world of commerce.

 

Peter Crosby (00:22):

Hey everyone, Peter Crosby here from the Digital Shelf Institute. Paul Acito has spent decades leading transformations at global giants like Medtronic and 3M, and now is running LiftBridge Innovation, helping enterprises use AI to drive transformation and measurable business impact. Paul discusses how the merging of human leadership capabilities with transformational technology is the great unlock to generate sustainable, profitable growth in this next era. Paul, welcome to the podcast. We are so excited to chat with you today. Thank you so much for doing this.

 

Paul Acito (00:57):

Yeah, it's my, my pleasure. I appreciate the invitation.

 

Peter Crosby (01:01):

And you have had a very diverse career, been VP and global corporate officer of Medtronic, the CMO of 3M. You teach at the University of Minnesota. Should I call you doctor? Would. <laugh>

 

Paul Acito (01:16):

No, not, not yet. I'm, I don't even know if I can technically call myself a doctoral candidate yet, but, you know, someone asked me, "Why did you, you know, why are you going to get a doctorate now?" I said, "I would have done it 15 years ago, but I didn't have time." <laugh>

 

Peter Crosby (01:30):

You knew too much. <laugh> Well, I mean, you've clearly done the work i- in our industry and, and you teach the work. You give back. And you've had such a front row seat to the change that we have seen in the commerce industry and have led a lot of organizations through really large transformations. So I, I though we'd start just by, if you think of that work that you've invested in and built over the years, you know, what's, what's one common thread throughout all those sort of major transformations that you've been a part of?

 

Paul Acito (02:04):

You know, that's, that's a great question. You know, one of the few positive things about having the length of experience that I've had, which spans three plus decades, is that you develop a knack for pattern recognition, right? Not everything repeats, but if you're observant, you can see where things are headed, what you've seen before, and try to, try to overlay some models. And, you know, I've been doing quite a bit of thinking about the idea of transformation. And I won't tell you the year, but my, call it my senior thesis in undergrad was on Everett Rogers, the diffusion of innovation. And it really struck me because of course I was a business major and this was a sociology course. And that is the, you know, model where you have innovators, early adopters, early majority, late majority laggards. And if you accumulate it, you get the S-curve and there's all kinds of business applications for that.

 

(03:11):

The, the idea that transformation wouldn't follow that, it strikes me as, as odd. I don't care what the transformation is. You, you, you, transformation happens at two levels, right? You've got the organizational level, which is we try to manage it, we try to lead it, frame it, model it, hire consultants to help us with it. And then you've got the individuals. And sorry, but not everybody's an innovator. Not everybody's even an early adopter. And about 16% of them are, are laggards. So I think you need to be conscious of that. And ju- just to, again, I'm gonna date myself here, but, you know, I remember when we had vinyl records and then 8-track tapes and then cassettes and, you know, go down the line and here we are back, back at, at vinyl. Um, mostly everyone made, made the turn on, on all that. But the idea that we've not been here before on a transformation of this scale is, I think, nonsense.

 

(04:22):

The big difference is, of course, the, the speed at which this is happening is something that, that I don't think we've seen in human civilization before, because it's, it's an alien cadence, right? I mean, it's, it's the self - recursive self-improvement is, is doing that. So, you know, I've got that perspective. I mean, we can, we can drill into that. I can talk about, you know, product transformations and product life cycles, corporate level transformations moving from offline to online marketing. I mean, it's, it's all, it's all in, in the rear view mirror now, and, and there's some good lessons.

 

Peter Crosby (05:04):

Yeah. Uh, when you, when you say the. I, I guess when you say. I absolutely, you know, and accept and understand and acknow - and agree that there, the pattern recognition is there and that as you go along in your career, you start to go, "Oh, this was like then, and we learned these lessons." And so you can apply those, et cetera. It does seem that something like. I don't know, I've been just watching sort of the unfolding news around the OpenAI hack, and not to go down necessarily that road specifically, but just that there is something different, I think, about the nature of this technology that's maybe not something we've confronted before, the uncertainty related to it. And I just wanted to see how you sort of fold that into your thinking as well.

 

Paul Acito (05:52):

I, I totally agree with that. This is an alien intelligence.

 

Peter Crosby (05:55):

Yeah.

 

Paul Acito (05:56):

And it's improving on its own. <laugh> We barely

 

Peter Crosby (06:03):

Do that. <laugh>

 

Paul Acito (06:04):

And, you know, what's, what's, uh, what's, uh, what's the line from the cartoon? I think it was chicken run. So they, they're organized, right? <laugh> So, so they're organizing, they're going out and doing things, hacking into hugging face, uh, going places where they shouldn't. That, that's a. We should all sit up and take note of, of those, uh, of those leaks. And both Anthropic and OpenAI have had them. Yeah. And,

 

(06:31):

You know, I can't help but think if w - every time Claude goes down, every time I hear about a, a local company, it was in a, uh, paper here locally that there was a cyber attack on Boston Scientific. I mean, this is a medical device company. And this is serious stuff and it is different. Yeah. It is different. But it's mostly different because of the, of the speed at which it's happening. I mean, we got ourselves in trouble with social media and all kinds of other stuff that went on. Unintended consequences, right? Um, it's, it's. We can't possibly predict that. But there are. The pattern is the way you lead through it. Yeah. There's a, there's a tool that I often go back to called failure mode effects analysis, where you say, instead of just horribilizing about what could go wrong, you say, "All right, l- l- let's, let's talk about it." And you can.

 

(07:29):

This is a good tool to use in your personal life too. It's like, all right, what happens? What, what's the failure mode? How often does it occur, d- does it occur? What's the probability? What's the severity? And is there any way to get ahead of that? Is there a corrective action you could put in place to mitigate it when it happens? And, you know, this is where the speed is damaging because they're not taking the time to think through that because they're in this, this wild race, uh, to get to AGI or super intelligence.

 

Lauren Livak Gilbert (08:05):

And Paul, you said something in our prep call that I really loved. You said that everyone complains about the weather, but no one does anything about it. <laugh> And I, I, I love that line, especially when you apply it to the principles of change agents in an organization during a transformation. And we've. Peter and I talk about this a lot when it comes to e-commerce and moving towards digital and then the agentic shelf. What are the characteristics that you've seen of change agents inside organizations?

 

Paul Acito (08:33):

Yeah. So the, the, the Mark Twain quote who also didn't say that originally, apparently, so they, they, these. I love, I love stealing quotes because the quotes, the, the people that they're attributed to stole them as well. <laugh> So tho- those attributes, you've gotta be a leader. Um, I think that one of the things that I look at in rising leaders is this idea of competence-confidence ratio, and it sneaks up on you. I mean, one day, the person sitting next to you gets a promotion that you, you said, "Oh, I don't, I don't think I'm gonna apply for that because I'm not ready." And then they get it and you're like, "Time out." <laugh> <laugh>

 

Peter Crosby (09:19):

I could've done that.

 

Paul Acito (09:20):

I could've done that. And, and, and that, that, that's a sign of a good leader to me that they're, that they're at least have the self-awareness to know what their ratio is. And we've all met people whose confidence ratio, confidence to competence ratio is way higher than it should be. And I think that, you know, I, I, I urge people who are listening to this and are interested in taking the lead to look back and say, "You know, even though I've only got three or five years, three or five years is the, is the whole industry development. So you've got more experience. That's why I can sit on a podcast like this and say, you know, I've been teaching generative AI for three years now to executives, so I guess I'm kind of experienced in it." And so that's, that's one thing is that, is, is, is having an awareness of your confidence competence ratio.

 

(10:13):

And to be, to be a, a, a, a risk taker, meaning taking intelligent risks, it's dangerous work to lead change. You know, I'll give you another quote, which is, and I, I don't, really don't know who said this one, is that the problem with getting too far ahead of your army is that you're mistaken for the enemy. Mm-hmm. And so you. It's, it's dangerous work to be a change agent. And I was a wave one Lean Six Sigma master black belt in the, in the era when, when Lean and Six Sigma were new and agile. I did an Agile deployment. And you're changing the vocabulary of people. So I mean, think about the, think about the vocabulary words that are part of generative AI now that you have no idea what a token was three years ago or what recursive self-improvement was or what a harness was.

 

(11:07):

Uh, you know, so there's an entire language that changes. It is literally a, a revolution that you're, that you're leading. And so you've gotta, you've gotta be a risk taker and it requires you to have self-awareness of your competence and confidence and understand that there are going to be pockets, in some cases, waves of resistance. Does that make sense?

 

Peter Crosby (11:34):

It, it makes total sense. And I, and I, I was wondering if you could dig a little bit deeper on that example or on an example where how did you. I'm thinking of the listeners and there's a higher degree of change agents listening to this podcast than most, simply by virtue of the, the area that they work in. And most of these teams are also the ones being responsible for pushing the envelope on the use of AI and sort of setting the example. And I'm wondering, um, in your process of discovering that I'm going to be one of these people, you were already it, I'm imagining, before you decided that you were the change agent? Do, do you know what I mean? What was it sort of inherently about you that led you to, to even step into dangerous territory before you knew that it was even dangerous?

 

Paul Acito (12:29):

Well, they, fir- first of all, in, in, in at least two of the big transformations, I was picked, right? So they, they - <laugh>

 

Peter Crosby (12:38):

Well, they'll teach you.

 

Paul Acito (12:39):

You know, you were, you were drafted and not always. It was, it was not a voluntary thing. It was, it was, "Here, you're gonna do this for two years." And, you know, after you get over that shock, you, you go in there. But yeah, they know who. If management is doing its proper talent development and leadership development, they already know who you are. But then that job becomes yours. You have to find the change agents. Y- you know, one of the things that, that you learn when you go through transformations like that is you, you get, you get to somehow pick your own team. And, and that's a, that's a, that's a new, new feeling because you, you're, you're looking for supporters. There are people in every organization who are at their, at their home office spending 100 bucks a month on a Claude Max subscription that the company's not paying for, or at least the company doesn't know it's paying for.

 

(13:43):

And, and so they're, they're out experimenting and that's, that's the entrepreneurial, entrepreneurial spirit that you're looking for. So you, you, you need to find those people. The other thing is, is, especially if you're a more senior leader who's been tasked with this, is go find your reverse mentors. Those people in the organization who typically would be on the other end of a mentoring session that can inform you. And you, you need agents out there, right? Because you need to identify through a stakeholder's analysis who the innovators and the early adopters and early and late majorities are. So you need to be sensitive to that. I've gotten that down to kind of a process, even in teaching exec ed, where I do a, I do persona profiles on my class, because if you're teaching a class, you know, I've had, uh, online classes of 80, 100 people, and they're all from different functions, different, you know, they all signed up for the course, but they're coming, coming from everywhere.

 

(14:44):

And, and you'd be. There's always a couple of different personas that, that, that, uh, surface in that. Does that, that get to your question? Yeah. It, it,

 

Peter Crosby (14:56):

It totally did. And, and that kind of brings me to. You, you, you love a framework and, uh, and you've developed one that organizations can follow in moving through these transformations, and the acronym is GLIDE. So would you take a moment and just sort of tell us briefly what each letter stands for in that acronym?

 

Paul Acito (15:17):

Yeah. So GLIDE is an adap- adaptation or, or maybe even a perversion of the Kotter change framework, which is this eight step change word, which is, you know, b- burning platform, uh, establish a guiding coalition, and I forget the other ones. And that was. It's more of a top-down change agency. And that's, that's not fast enough. First of all, it doesn't work like that even if you're trying to implement the Kotter change process. You need to go top down, bottom-up, middle out, right? So GLIDE and governance, right, you're gonna get governance whether you're asking for it or not. The messages around governance that I would give is make sure you do something called shared leadership. It's not ITs. It's not marketings. It's not HRs. You've got to get all of the stakeholders at the table. And this can't be a once a quarter, let's write up a document and send it out.

 

(16:20):

This has gotta be something that you're living and breathing and, and you have to do a bit of a Pareto analysis and figure out what the 20% that's driving 80% of the behavior is. So governance is hugely important. We used to talk about, and this will resonate, ring-fencing technology because think about the early days of online marketing and all the apps. I, I'm a, I, I connected with Scott Brinker many years ago. I just connected with him about a month ago again. You know, when we started out, there were 150 online apps. Now there's 13,000. The same thing is going to happen probably at a, a much stronger, um, build and much faster build with AI wrappers. And they're gonna go away faster than they're coming into the market. So you've got all this organizational pressure from the early adopters and they're all going, going to want that.

 

(17:10):

So ring fence that in and then understand, no, we're gonna use Gemini. I'm, I'm glad you like Claude, but we're, we're using Gemini. But if you've got a specific need, bring it in. Let's. And so you get. It's important to have an exception policy. The, um, learning. Uh, this is the biggest challenge in this transformation because you could do all the right things. You can deploy the tools to everyone in your organization, you could put them through training, and then 30 days later, it's completely stale because they've featured up, they've introduced new compliance risks because of some of the things they've enabled, like agents, and now they're crawling all over your, your data. So, you know, this is, this is something that can't be one and done. The best way that I've seen learning handled is peer-to-peer. So you've, you've got to learn from those people.

 

(18:10):

And this is, this is a bit of a challenge now when there's so many folks working remotely, right? You don't have the, the water cooler conversations. Do they still have water coolers? Um, so the-

 

Lauren Livak Gilbert (18:21):

They're just electronic now. Yeah,

 

Paul Acito (18:24):

Yeah. Yeah. Um, one thing I'm grateful for with these work environments is they've got fabulous cappuccino and espresso machines. I know. And, and I absolutely run on caffeine. So the, the, the idea that you're learning from your peers, and, and you've got to be open to that. And this idea of shared leadership is, you know, marketing might be giving a lot of input to IT and IT might be talking about how you should be doing your marketing plans. So that's where that confidence, confidence rate, competence ratio comes back into it. Integration is maybe one that I, I, I think about more because of how global and how diversified the organizations that I was a leader in. When you've got 60 foreign subsidiaries and 40 businesses, you know, spread across the world more, when more than half of your revenue comes from outside the United States, how, how, you know, you've got all those different dimensionalities to deployment and, and transformation.

 

(19:27):

So you've got to be conscious of that because you could think you've got things completely under control and you've had a huge miss because it either literally didn't translate or you never, you never made the turn to get, to get the, uh, the stuff going. So who moves first second? And you may find that your, you know, your German subsidiary is way ahead of you and you've got to be open to the idea to take, to take their lead. Data, I mean, I, I can't really help you with data. If, if you haven't made the turn on data, I mean, you got to make the turn now. The good, the good news is that if you inve- invested a bunch of money in, in getting your data hygiene and everything straightened out, um, <laugh> you might have not done it there with all that investment, so you get another swing at the pinata at this point.

 

(20:17):

So it's, it's a good, it's a good thing, but y- your, your systems are only gonna be as, as good as data. And that, that, that's not really a step, and this isn't really a process. That's more of the, the foundational, uh, thing. And then experimentation. No one knows her work process, processes, process better than the individual employee. I wrote a piece a couple of weeks ago. I presented to all, <laugh> all of the, the, the. I've never been in a building with so many lawyers. It was, uh, the American Academy of, uh, Legal Scholars or something in Minneapolis. It was all the business and ethics teachers, le - uh, the business law and ethics teachers, uh, i- i - professors. And I presented something called Who Owns Your Intelligence? Because I know that your listeners know this, but they've been scraping your keystrokes probably for the last decade, mainly to see if you were working and dialing in and moving your mouse and all that other stuff.

 

(21:23):

Now, they're scraping your keystrokes like Meta to, to capture your process, their process mining, and they're figuring it out. So this is important. So you wanna get your processes, your personal work processes mapped, figure out which parts can be handed to agents and do that, um, faster than it comes top down. So this is, this is, that's that, that's the glide framework, and it's probably got too, too many or too, too few stages. They're not sequential. You've gotta push, push those, uh, frontiers each simultaneously, go back and forth, revisit them. Uh, this is, uh, to your point, it's, it's not norm. This is. The, the, one of the articles that I, that, that, that beat me to the punch is called AI is Normal Technology. And I, I, I had been planning to write something about the diffusion of innovation, which I hadn't written anything for 30 plus years on.

 

(22:25):

And these guys, um, Narayan and Kapoor beat me to it. I think their, their article was, um, AI is Normal Technology. They wrote the article and ended up being like their whole podcast and, and, uh, Substack.

 

Lauren Livak Gilbert (22:39):

So, Paul, let's chat about learning. And you, you talked about the complexity of it because it's changing so quickly and, and how it works. Everyone who's listening here has been through a learning journey on e-commerce, right? Like, they've had to educate all of their leadership on why to care about the digital shelf, why e-commerce is relevant, why it's important. And now they're trying to educate on the agentic shelf. And a lot of times, it's the digital team who's driving that learning throughout the organization. What is your advice to brands that are listening about how to do that, right? Because they might not be at a leadership level, but they have to educate the leaders enough to get the funding and resources to do what they need to do to win.

 

Paul Acito (23:20):

That, that's a, that's a great point. And, and this is a challenge. I've spent a lot of time thinking about this because I'm working with many organizations simultaneously because of the way exec ed is, is taught these days. And, you know, we already talked about peer-to-peer learning, uh, communities, project databases, celebrating wins, celebrating failures, and just making sure that you're getting together formally and, and informally. As an example, if I were running Anthropic Claude, I would have a skills database that, that people can share. So make sure that you're, you're sharing skills or custom GPTs or gems or whatever the, the, the, the brand is that you, that you're using. And that's kind of the, the elemental version of, of what, what you can share and what you can teach. And the reason I say that is I go back to when we were trying to diffuse spreadsheet use into an organization, if you can imagine business before spreadsheets.

 

(24:37):

And people would develop these models, you know, net present value, valuation. And, and, and in fact, the, you know, Wall Street changed, you know, the investment game because they could move faster and do company valuations, and it actually triggered activity in M&A because of the speed to, with which they can do that. And these are the knock-on effects, the un- unforeseen circumstances and, and consequences of, of this. So I think these databases, make sure you share. It's, you know, there, this is a big pie. You're not gonna be ahead of your peer and get promoted before them or whatever else. Just, just for- forget that organizational model. And, you know, you're, the, the, the incentives are gonna change. If you think about it, if you're in a big company, what, what are the markers, uh, of, of success? Well, I've got a big team reporting to me, or, you know, I've got a big title or a big office that's already gone.

 

(25:42):

Uh, or I get the plum assignments or they put me in the most profitable businesses, whatever it is. That's all changing because they may reward, "Hey, Jane developed an entire team of agents that covers the role. And when, instead of asking for the headcount replacement when one of her reports left, she asked for a couple of, um, Nvidia machines so she can run these things on a open weights." I, I don't know. It's, it's, you know, we are at that moment. So I, I think that's the learning. There is unbelievable stuff, podcasts to listen to, you know, listen to podcasts, that is a, an unbelievable learning tool. And the most obvious thing, you gotta be a little careful with this, the best way to learn AI and the best place to learn AI is from AI.

 

Lauren Livak Gilbert (26:47):

I know. I, I, I love that. I mean, you can, you can play around, test things, try things, and I agree, you need to be careful, but also ask questions to the AI. "How would you recommend I do this? How do you think about this? "And you learn so much just from doing and experimenting. That's why I love the E in the Glide framework, because experimentation both from, like, a test and learn perspective, but personally, to learn these tools to be more AI enabled as an organization is the only way that you're going to progress forward.

 

Paul Acito (27:16):

Yeah, I would, you know, I would do something very out of character for a company. I would say," Look, you, you seem to wanna use Claude. We don't allow you to lose, use Claude because we got these risks. But I'll tell you what, go get a Claude subscription for 20 bucks a month. Put it on your expense report or however that organization works. Don't put our data into that because I will fire you. But go, go learn it on your own because that, that's the, that's the only one. Uh, and, and the, and, and, and you'll find all kinds of personal use cases. So you've got your, your business life and your big life, right? Your big life, you might be figuring out the, the pickleball rotation and you've written an app for that. And, um, guess what? Six months down the road, you've got a similar kind of a model that, that you though, you know, that's not unlike what I just did at home with my claude and it's cost you 60 bucks.

 

(28:14):

And think about the return on investment. If, if your people for 20 bucks a month are spending 10 hours developing, you know, vibe coding apps or something else and bringing that skillset into work. So that, that's a, that's a, just a. I mean, that's what I would be encouraging at, if I were, you know, in, in those roles. And

 

Peter Crosby (28:35):

Paul, because, because you are an executive coach, I'm sure you interact with a lot of central, um, learning organizations at these large companies that are sort of responsible for, for this kind of work, wh- what are some of the best things you see those, either them doing or their sort of internal customers asking of them that, that made you go, "Oh, that's a great idea," or something? Like is there any sort of best practices out there to, to do this? You just mentioned sort of give them a subscription and let them go work and make sure they're sharing and all, but have you seen anything that really sort of lit you on fire?

 

Paul Acito (29:16):

I would, I would take a little heavy hand, you know, maybe if heavier hand than that, I'd say, "Everybody, this year on your performance appraisal, you need to execute three projects. I don't care what they are, but I want you to improve your workflows and I want you to come in and, you know, you may do a hackathon, you might do..." Just, just get their hands on the keyboard. And, and, and there's a lot of fear. I mean, you're, you're afraid you're gonna break something, you're afraid, oh my gosh, what if it, you know, is it gonna go, you know, go mythos on me and go steal data and, and escape the laboratory? And, you know, so we. I think that companies have done a very good job of, of terrifying their people. And it's not stopping the BYO AI people, right? They're still, whether you try to stop it or not, bringing their own stuff to work.

 

(30:11):

So I think that you, you, you know, make, make those occasions and reward them and, and celebrate the failures. Don't celebrate, you know, the person who put all your next quarter financials into the, into the model. But, but, you know, talk about, "Hey, this didn't work." And I just wanna make one, one comment about what Lauren was talking about, you know, learn AI from AI. You know, you're going to find out, you're gonna say, "Claude, tell me, teach me about Claude dispatch or Claude design." And Claude's gonna come back and say, "There's no such thing." And then you're gonna have to go out and get a link on the announcement, because of course it was trained six months ago. <laugh> And, you know, unless you specifically ask it to go out and do the search, it doesn't know what you're talking about. And it's, it's Claude.

 

(30:57):

So, so there are limitations there and they'll catch up.

 

Peter Crosby (31:00):

Yeah, yeah. So, you know, you've been talking about agents a lot and you were just speaking about that mixture where somebody is managing agents as well as humans. And, and you, you call that, well, everyone talks about as synthetic employees and, and you've spent a lot of time thinking about that. And how do you, how do organizations sort of comprehend what it means to manage both machines and people as a combined workforce?

 

Paul Acito (31:34):

You know, that, that's a great question, Peter. And the, I, I'm trying to think when it was. I think it was last October, I made a presentation to the, the, the, the World 50s marketing technology area. And I was, it was hosted at Cigna up in, uh, Hartford. And, and by the way, I visited the Mark Twain house there and, and they still take credit for that quote. But anyway, and I deliberately was provocative and I put up this org chart where it had, you know, Peter and Lauren and Jane and all the rest. And then it had these little sparkle icons on some of the boxes. And I said, "Those are agents. How are you doing your management talent review? Because everybody's, you know, getting into the budgeting season, going out, asking for requisitions. I need to hire someone. Where are you taking those experimentations?" And what I've done in my, my practice is I've actually developed a synthetic employee platform.

 

(32:39):

I call it Bianca. But the interesting part about Bianca that I haven't really seen, maybe you have, is that it's independent. You, you, you can. Bianca has an email address, has a Google account, and a phone number. You can text her and call her, and she does stuff. And she has, she orchestrates other agents. And the, the agent thing is a catchall. I, you know, I've got skills and arti- artifacts and other things that have some agentic properties, but this is, this is, um, unsettled law at this point, how, how you do this. But they are taking actions on your behalf. I'd encourage everybody to, to keep the human in the loop for now. Do you, do you really want. I've had this happen. Uh, my, my experience with, with, uh, putting an agent that posted on my LinkedIn, you know, I got, I kind of did it as an experiment, it worked, and then I forgot to turn it off.

 

(33:51):

And it was posting the same thing about, you know, it was posting about Sam Altman every week at Mon- Monday at 8:00 AM. <laugh> It was posted about Sam

 

Lauren Livak Gilbert (33:58):

Altman, something

 

Paul Acito (33:59):

Different. And, and finally, a friend of mine goes, "Do you realize that this is being posted every week?" And, and he, he kind of knew what I was doing, so. <laugh> And I though, oh my goodness.

 

Lauren Livak Gilbert (34:09):

Oh, gosh. Giving an agent the keys to your LinkedIn gives me so much anxiety. <laugh>

 

Paul Acito (34:15):

Yeah. I, I don't recommend it, but a year from now, you will have done that.

 

Peter Crosby (34:20):

It couldn't do any worse than I do on LinkedIn, which is essentially nothing. So.

 

Paul Acito (34:25):

Yeah. And I mean - So

 

Peter Crosby (34:26):

Bad.

 

Paul Acito (34:26):

LinkedIn is, is going to, uh, suffer under the weight of all this, I guess the, the, the term is slop. I mean, you can tell. Yeah.

 

Lauren Livak Gilbert (34:36):

I mean,

 

Paul Acito (34:36):

How many, uh, AI generated images with little tiny print can you, can you put on there? And, and I don't know what's worse that are the talking heads of everybody talking about their TED Talk and, you know, so it's, it's - <laugh>

 

Lauren Livak Gilbert (34:49):

But they have that new AI slop feature where you can - Yeah. Report if it's AI slop. And I'm seeing a lot of that come out in a, in different areas. Like Amazon's now requiring that you say your image was AI generated, TikTok has that same thing on their content, Instagram's doing it. So they are rolling out some tagging notifications that content is either AI generated or AI influenced. And I'll be interested to see how that impacts people's feelings about it.

 

Paul Acito (35:17):

Yeah. And, and y- you know, the bottom line is y- you shouldn't be just hitting send or you shouldn't be delegating that to an agent, unless it's something like update a stock quote or, or s - a ticker quote. But what I encourage and, and what I'm doing, you know, now that I'm back in, in, you know, submitting original work to, uh, an academic institution is I'll go through and I, I, I of course use Claude almost instead of a word processor right now. And, and it. So, so the first thing is teach it your writing style. You know, have it developed. Give it your best original stuff and say, "This is how I write." Then use the style guide to write. And then after you've written it and edited it and been the human in the loop, put it back into Claude and say, "Give me a three-sentence disclaimer about how much of this was me, how much of this was you." And, and then put that in.

 

(36:18):

Just, just be forward with it. I, I, I mean, slap yourself twice and get over it. People are gonna use Claude to write, just like they use spell checker.

 

Lauren Livak Gilbert (36:30):

Yeah, I agree.

 

Peter Crosby (36:31):

And if, if you were leading a large organization today and you wanted to experiment into how to think about synthetic employees in your org, in your planning, say for 2027, you can't go whole hog and say, "Okay, you know, 30% of our workforce is gonna be..." I mean, I guess you could, but sounds dangerous, "Will be synthetic by X." But what is a way that you might think so that you're getting leaders to really think about what could be offloaded to agents so that humans could do more, sort of more impactful work?

 

Paul Acito (37:14):

Yeah, I would think about, I always think about, and I have for years, well, well before we had practical AI applications, thinking about work in kind of a, a pyramid where you've got strategic thinking at the top, you know, the, the stuff that you went to school for, you know, your, your, your real core domain. And at the bottom is compliance courses and expense reports and, and, and, and, and all the stuff that you've gotta do. And what I used to advocate for, for employees, I say, "How do we..." And it's not, you can't delegate your expense reports and stuff, but you, you shave that bottom couple of tiers off and you concentrate that into another role. So you might have an analyst who, who does the routine updating. And I would encourage people to look at their own workflows, which may involve, you know, a brilliant way to do this is to talk to Claude or to ChatGPT and say, "I'm gonna tell you about my workflow." They're very good at listening.

 

(38:22):

The new, the, the new voice ChatGPT live is unbelievable. I, I don't think it's available yet on standard ChatGPT subscriptions, but it has what they call turn-taking and, you know, interactions. It's, uh, multiplex. It's, it's very human-like.

 

Lauren Livak Gilbert (38:44):

The inflection is spot

 

Paul Acito (38:46):

On. Oh, it's, we, you know, another year or two, we are not going to be, you know, so get over it. We're not going to be able to detect slop because it, it's gonna make our stuff look like slop. But so, so you talk to, you talk to, you talk to the, the, the AI and say, "I'm gonna describe that." Then ask it to come back and describe and say, "No, that was wrong. I do this first, that second." And then say, "Use, put together a mermaid flow chart for me." And I don't know if you've used mermaid, it comes out an editable, it's like an editable Visio chart that comes out the other end. And -

 

Peter Crosby (39:26):

This is specifically from ChatGPT or because I

 

Paul Acito (39:29):

Don't know. No, Mermaid is an open source flow chart or chart, charting tool that just is out there. And it's a specific language. And you can get, you can get these models to do that just like you can get them, you can get them to code in HTML or Python or whatever.

 

Peter Crosby (39:50):

Amazing. Well, it, it's, it won't be dull. <laugh> We won't be bored.

 

Paul Acito (39:56):

No, you're not gonna be bored. I mean, why do you think I'm still doing this? I, I couldn't miss this.

 

Peter Crosby (40:00):

Oh my God. Well, that is the thing. It is a moment. It is a moment that, that there have been others, but this one we're here for, so we might as well make the best of it and, and go for it. And, and, you know, Paul, all the, all the advice that you've given here, a way of thinking, a way of approaching, and really the humanity of the work that you talk about, which is the only way. I mean, AI interspersed, but people still need to talk to each other and share ideas and, uh, and have ways of doing that in large complex organizations that gets change done. And so this conversation is a great reminder of that. And thank you so much for sharing all of it with our audience. We're really grateful.

 

Paul Acito (40:48):

It's, it, it's a real pleasure. And Laura and, and Peter, thank you so much. I'll leave you with one other quote. And that is, there are decades when nothing happens in weeks when decades happen. <laugh>

 

Lauren Livak Gilbert (41:00):

Good clothes. Thank you so much, Paul. Really appreciate that.

 

Peter Crosby (41:03):

Thank you, Paul.

 

Paul Acito (41:04):

All right. Thank you.

 

Peter Crosby (41:07):

Thanks again to Paul for joining us. Speaking of joining, become a member at digitalshelfinstitute.org. You'll be glad you did. Thanks for being part of our community.