91制片厂

00;00;00;12 - 00;00;16;07

Geoff Nielson

Hey, everybody. My name is Geoff Nielsen, and this is Digital Disruption. Joining me today is Connor Grennan, chief AI architect at NYU stern and bestselling author. Connor, thanks for being here today. Yeah, thanks for having me. So I want to jump right into it. You've got a few hot takes when it comes to AI.

00;00;16;18 - 00;00;28;26

Geoff Nielson

The first one that you know, we can jump into around adoption is about the importance of change management. And this having kind of a more prominent role than even the technology, you know, can you maybe walk us through. You know what you mean by that?

00;00;28;28 - 00;01;07;11

Conor Grennan

Yeah. It's a very strange thing. This is a technology, and I mean generative AI, ChatGPT, etc.. That doesn't really act like a normal technology. This is sort of a software that acts more like a human than a software. And so what I have, what I've found over the last, two years of, of really working intensely with this, both inside NYU and also working with a ton of very large companies, is that the technology is not nearly as important as you say, as, as the behavior and maybe the the best analogy here is usually when you're doing a digital transformation of some kind, you're using, you know, you're going from some old system up

00;01;07;11 - 00;01;33;25

Conor Grennan

to, I don't know, Salesforce or a CRM or moving over to Excel or whatever it is. And the way that you do that usually is to teach people how to use that software. And there's a learning curve to it, and you get rid of the old software, and then everybody's eventually using this new software. And I think the, problem is that that's how everybody was handling something like generative AI ChatGPT or Microsoft Copilot or whatever the tools are that people may be using.

00;01;33;28 - 00;01;53;23

Conor Grennan

And, it turned out that that didn't work, which is why adoption rates tend to be very, very low. So instead of comparing it to something like, you know, Excel or a new CRM system or something like that, it's actually much more like a, like a treadmill, right? Where it's not actually about learning how to use this treadmill.

00;01;53;23 - 00;02;12;21

Conor Grennan

It's actually just about the behavior. So you could put a treadmill in every house in America, but you're not going to see heart disease really lower. And the reason for that is, is that, again, it doesn't really matter about the technology as much as it matters about the behavior. And that's because this technology like generative AI, it doesn't really require you to learn anything.

00;02;12;21 - 00;02;26;01

Conor Grennan

There's not really, you know, better ways to prompt and there's not really oh, you need to understand AI or machine learning or anything like that. You really just need to have a conversation with it. And that's where we start launching into what's so different and how we treat it differently.

00;02;26;01 - 00;02;37;17

Geoff Nielson

Right? Right. And so you've, you know, you've talked to no shortage of organizational leaders about how to improve adoption of AI. And, you know, one of the things you're not recommending is use cases, right?

00;02;37;23 - 00;02;47;29

Geoff Nielson

Which is kind of the conventional wisdom. You know, everywhere we look, it's find your I use cases. What's, you know, what's kind of the, the issue there and what advice are you giving them instead?

00;02;48;08 - 00;03;05;24

Conor Grennan

Yeah, I know that's a very strange thing to say, right? Because again, it's really the best way to learn a new technology or really to learn anything is through use cases. You want to give examples, all that other stuff. There's a reason why I don't do it that way. And the only reason is really it doesn't work.

00;03;05;24 - 00;03;06;14

Geoff Nielson

Yeah. So what

00;03;06;14 - 00;03;11;04

Conor Grennan

I hear people saying and people I really love and respect and everything is, hey, look, you just got to practice.

00;03;11;04 - 00;03;31;17

Conor Grennan

You just got to use it. All that kind of stuff. Why doesn't that work? Well, it doesn't work for the same reason we say, oh, just, just eat less and exercise. Just get on that treadmill. That doesn't change behavior. You can tell people that, but it doesn't actually work. So when you tell people just to use it without recognizing why our brains are pushing back on this and everything else, it's not going to get us anywhere.

00;03;31;17 - 00;03;47;21

Conor Grennan

So what I found with use cases very early on, actually, was that I would go into, I don't know, like a big healthcare organization or something like that, and they would say, hey, bring us the top ten use cases. I was like, great idea. So I would Google ten use cases, say, here's how you use it. And we'd walk out of that workshop.

00;03;47;21 - 00;04;07;27

Conor Grennan

Everybody's happy, super excited, everything. And then I would check back in with them maybe a month later and a strange thing would be happening. Right? Which is I would say, how's it going? They're like, oh my gosh, amazing. Those use cases are tremendous. They've really helped speed up this and this and this and I my reaction always be but but you're you're just like I didn't know I don't know anything about healthcare.

00;04;07;27 - 00;04;26;05

Conor Grennan

Like I googled those use cases. And so what was happening is I started to talk to people who really understood the brain and everything else was that the brain loves templates, right? We love to sort of understand, hey, here's the tool. What do I use it for? Right. So and if not, then the brain is just going to try to understand like, well, what should I be using this for?

00;04;26;05 - 00;04;41;23

Conor Grennan

It's like when we switch out, you know, the flip phone for the iPhone. But when I did that with my father, who was very old, he just kept on using it like a phone because he didn't realize that this isn't replacing your flip phone. An iPhone is replacing a laptop. It's replacing how you order Ubers. It's replacing how you watch TV.

00;04;41;23 - 00;05;03;24

Conor Grennan

It's replacing all that. And so when we just use use cases, your brain is very happy because your brain wants templates, your brain wants to understand what this is. Your brain wants to pull this thing off the shelf to use it in those specific moments, instead of using this to augment everything you do. So what I use instead of use cases is trying to break down what the brain is doing.

00;05;03;24 - 00;05;17;04

Conor Grennan

Why the brain is doing it, and a whole new way of thinking so that people can come up with their own use cases. And then of course, you need use cases for demonstrations. You need it for inspiration. But ultimately people have to come up with their own use cases, not have them dropped from the outside.

00;05;17;04 - 00;05;30;28

Geoff Nielson

Right? Right. And I think that's, you know, that's one of the inherent tensions of organizational use of AI right now. Right, is like for managers and leaders, there's a real, there's a real temptation to just want to template ties everything.

00;05;30;28 - 00;05;46;22

Geoff Nielson

Right. I'm going to tell you, how are you going to do this versus, you know, employees saying, this is how I'm getting value out of it, you know, how do you see that evolving? Is this something that employees are really going to continue to lead to be successful is there's a is there a role for leaders to be able to drive this more?

00;05;46;26 - 00;05;49;07

Geoff Nielson

What does that look like over the next few years?

00;05;49;10 - 00;06;05;22

Conor Grennan

Yeah. It's such a good. It's such a good way of saying I'm going to steal the word template ties. I love this word. But you're absolutely right. Right? I mean, like, people just want to know how to use this. And by the way, usually in the history of technology, it's pretty easy, right? The flashlight replaced the candle so people aren't walking around like Bennys or Scrooge anymore.

00;06;05;28 - 00;06;24;12

Conor Grennan

But that's because your brain knows exactly what it's replacing. You see, for example, isn't replacing one thing in the same way the iPhone wasn't replacing just talking on the phone to your grandma, right? It's replacing a lot of things, which means it requires a change in behavior. And I absolutely think that leaders have to play a leading role in this.

00;06;24;12 - 00;06;42;14

Conor Grennan

I think that the problem is that often we're saying, well, let's just run a pilot project and see how that goes, and we'll go, well, but the pilot project doesn't tend to lead to the next thing, because people are excited about that use case, it's sped things up, etc.. So what I talk about with leadership is I think it's really critical that leaders understand this.

00;06;42;14 - 00;06;57;12

Conor Grennan

So when I, I used to go out and just work with teams and, you know, 200 person digital marketing team and it would be great, right? I mean, they would come out, they'd be using it all that kind of stuff. The problem for organizations and I work with kind of entire organizations now is that it doesn't really hit revenue.

00;06;57;16 - 00;07;15;00

Conor Grennan

Right? So in other words, like those people that you trained up will turn their eight hour day into a six hour day, but that doesn't hit revenue, or they'll just use it to be more impressive than their colleagues when their colleagues aren't using it. But that doesn't really hit revenue. And so the critical aspect here is that you have to train senior leaders first for a couple of reasons.

00;07;15;00 - 00;07;31;14

Conor Grennan

First of all, senior leadership has to understand what this is. I think too often they don't even really understand, you know, how to explain this to clients or how to sort of understand how this is going to impact every department. So you have to do the same training with leaders so they understand how they'll use it in their workday.

00;07;31;18 - 00;07;49;22

Conor Grennan

That's number one. Number two is the best leaders I find are the ones that go out there and they're vulnerable about it. They say, hey, listen, we're all learning this together. This is not looking. We're not looking to replace everybody with AI here. I should go through you, not around you. I has to augment something. And what it should augment is your best people.

00;07;49;22 - 00;08;09;21

Conor Grennan

It shouldn't just kind of come in here and just operate in a vacuum. But the product will be garbage if it does. Right? So. So leaders, first of all, have to understand what this is, how they use it in their workforce, and also how they can understand it so that they can talk to their workforce, but more importantly, they have to understand what a new benchmark is for an eight hour day, right?

00;08;09;21 - 00;08;28;17

Conor Grennan

I mean that the big problem is that they just know that this will speed people up. But once you have your line managers in leadership, essentially head of HR, head of legal, head of whatever, then they'll understand, oh, our expectation for a working day for somebody in one of our departments was this. But now with generative AI, it will be this and it will condense.

00;08;28;17 - 00;08;55;05

Conor Grennan

And so it hits, so they can understand how to set new benchmarks for their entire organization. And I guess the third thing is around talent. When leadership doesn't understand what this is, it's talent evaluation gets really skewed, right? Because if you're training a whole organization, what you see, and I've seen this a lot, is that, you know, you have your people who you know are the best in the business at something, but all of a sudden you have these kind of young up and comers who are suddenly outperforming them.

00;08;55;05 - 00;09;13;21

Conor Grennan

You're like, well, maybe that person should be the manager, but what's happening is that that younger person or that new person or whatever is just, you know, using generative AI and all of a sudden their work product is much faster, much better, etc. and it's very hard to understand how to reward talent and how to promote talent. It's just a huge problem.

00;09;13;26 - 00;09;33;00

Geoff Nielson

I'm I'm curious about that last point, Connor, because, you know, as you said, it's almost like it's almost like it's providing a super power to the employees that have figured out how to use it properly. And, you know, when it comes to rewards, when it comes to leaders trying to figure out how to evaluate this, like, does that ultimately matter?

00;09;33;00 - 00;09;50;15

Geoff Nielson

Do we need to separate the AI from the person if we've got, you know, a cadre of employees who have figured out, like they've got another tool in their toolkit, you know, do we need to care that it's AI? Is there risk to this? How should we be thinking about that? I think this is such a great question because it

00;09;50;15 - 00;09;56;29

Conor Grennan

goes to the point. One of the points, and I think one of the underlying things is, you know, is AI cheating, for example, you know,

00;09;56;29 - 00;09;57;17

Geoff Nielson

Yeah.

00;09;57;17 - 00;10;03;12

Conor Grennan

use it? And I think. No. Right. I mean, this isn't it's not high school. You're not getting graded on your, on your paper.

00;10;03;12 - 00;10;25;18

Conor Grennan

Right? I have kids and they can't use generative AI for writing papers, but they can use it for learning biology better. So it really depends. Right. But I think that, that most importantly, first of all, you have to understand that there are actual, you know, laws and limitations around this. Like, you can't just produce something that's AI, either in imagery or video or even text, to be honest, and just put that out into the world of yours just because you can't copyright it.

00;10;25;18 - 00;10;46;25

Conor Grennan

That's a that's a legal issue. But beyond that, we should have everybody in the organization with guardrails in place of course using this. And why is that? Because it is going to augment what they are good at. Sometimes the example uses, if you put me up against a marketer and you said, okay, in 20 minutes, both come, both of you come up with a new idea for a shoe company or something like that, right?

00;10;46;27 - 00;11;09;28

Conor Grennan

Yeah, I would produce something really awesome, even though I'm not a marketer. Right. Because ChatGPT would help guide me and it would be amazing. And after 20 minutes it would be incredible. But the marketers work product would be ten times better than mine. Why? Because they understand what quality looks like. The Henderson. It's sort of like when you say, hey, write a poem and you read this poem by ChatGPT, you're like, this is great, but a real poet would be like, that's literal trash.

00;11;09;28 - 00;11;24;02

Conor Grennan

Like, that looks like poem trash, right? Because the person who actually has the brain that this, this tools are going to augment, understand how to guide it, understand what quality looks like, etc. and to not have your people using that as kind of an Iron Man suit,

00;11;24;02 - 00;11;25;25

Geoff Nielson

you're really just shooting yourself in the foot.

00;11;25;25 - 00;11;44;06

Geoff Nielson

This is why it has to be completely spread throughout your whole organization, right? Right. And I think so too. And I love the analogy about the The Iron Man suit, that that's exactly how I envision it. But you used one word I want to come back to, which is guardrails. Right? That this is something, you know, you know, I talked to a lot of, you know, IT leaders.

00;11;44;06 - 00;12;02;00

Geoff Nielson

And, you know, one of the roles we talked about for it is, is, you know, this notion of guardrails and it's it seems like we haven't necessarily found the balance in terms of, you know, what's too much, what's what's too little, creates a risk, you know, what are you seeing? What's what's the most important role for guardrails?

00;12;02;00 - 00;12;09;25

Geoff Nielson

Who should be doing it. How do we you know, what principles can we use to make sure we get it right? It's I mean, I think this is really

00;12;09;25 - 00;12;19;18

Conor Grennan

tough, right? Because because and to your point, I think it's very easy for organizations, especially when you're in, like a regulated industry or something like that, to just shut it

00;12;19;18 - 00;12;20;18

Geoff Nielson

down. Right. Yeah.

00;12;20;18 - 00;12;26;01

Conor Grennan

too it's too risky. And that, I mean, it's unbelievable how risky it is to just shut it down

00;12;26;05 - 00;12;26;24

Geoff Nielson

Yeah.

00;12;26;24 - 00;12;33;18

Conor Grennan

your organization and it could be an existential threat to your organization if everybody else is augmenting their organization and differentiating.

00;12;33;20 - 00;12;57;12

Conor Grennan

So I would say the first thing when I, when I work with organizations, anyway, the first thing is I like to help them understand what is this thing. Right. Because I think a lot of people, when they say, oh, we can't upload this or this or this, just helping them understand what's actually happening. When you upload something now, you know, if you upload, for example, information about yourself or something like that, it's not like your competitor is going to come and swoop down and grab it off the table in Starbucks.

00;12;57;12 - 00;13;19;04

Conor Grennan

It's not like a lending library. It's not like I upload something and somebody could say, what did Connor Grennan sort of talk about in his when he was talking about personal finances? It doesn't it doesn't work like that. The tool doesn't work like that. However, there is a real, you know, cautionary tale around uploading proprietary data. And but that's first of all, that's mostly around legal issues, right?

00;13;19;04 - 00;13;36;23

Conor Grennan

I mean, like, you can't you don't want to violate third party agreements with people and you don't want if you have, you know, agreed in banking or something like that, that you're not going to share information, then you'll be violating that agreement or in legal, you know, around health, around education, things like that. You literally can't share it because you'd be breaking the law.

00;13;36;26 - 00;14;04;01

Conor Grennan

So those are the guardrails number one and two, right. Are you violating legal agreements, third party agreements, things like that. But then you get into, I think what you're talking about with it. Right. Like, well, what about code? Can we put that up, all this other sort of stuff? You just have to have guardrails in place. And by the way, these guardrails don't have to look wildly different from other types of guardrails you have in place around, hey, you know, like, I wanted to put this code up, you know, on some kind of open source place where people can, you know, on Reddit or whatever, and people can help you.

00;14;04;03 - 00;14;17;29

Conor Grennan

You're not going to want to do that either. So I think there are guardrails that you can sort of big guardrails you can put in place, and then you start narrowing it down for exactly what you do in your organization. It's something to be taken very, very seriously. But I think that the bigger risk is just saying,

00;14;17;29 - 00;14;18;16

Geoff Nielson

saying, you know what?

00;14;18;17 - 00;14;40;00

Geoff Nielson

We can't trust anybody. Yeah. It that to me is the huge risk. Yeah. No, that that that makes complete sense. And, you know, I'm curious, Connor, when we talk about these guardrails, this pace of change, some organizations taking more advantage or less or regulation, you know, one of the big questions that everybody's trying to figure out in their crystal ball is like, who's going to win from this?

00;14;40;05 - 00;14;59;13

Geoff Nielson

Right? What types of organizations are going to win? And, you know, when you look at the disruption factor, you know, looking across startups, looking across kind of these, you know, tech giants and megacorp and looking around kind of middle sized organization, you know, who do you see is coming out ahead? Who do you see as being at the most risk?

00;15;00;17 - 00;15;14;04

Conor Grennan

I think that, you know, that can kind of be broken down to a couple of different big groups, right? One is, you know, which tech companies are going to win. I like I'm all for, you know, promoting small startups, things

00;15;14;04 - 00;15;14;25

Geoff Nielson

that. Yeah. You

00;15;14;25 - 00;15;25;05

Conor Grennan

want to be careful when you're talking about your data. If you start, you know, trusting your data to a smaller organization and then they go out of business, you know, where is that data gone?

00;15;25;05 - 00;15;40;13

Conor Grennan

Where does that live? I really think, and I hate to say this, but here we go. Right? I mean, we have like the OpenAI's, the Googles, the anthropic, the metas, the AI, the Amazon and now kind of Amazon a little more. But you have these big, very trustworthy organizations, trustworthy,

00;15;40;13 - 00;15;41;11

Geoff Nielson

by the way putting in there.

00;15;41;13 - 00;15;42;28

Geoff Nielson

Sure, sure. But you

00;15;42;28 - 00;15;56;15

Conor Grennan

know, but these companies that have that are building this out because they want you using their digital cloud, things like that, that feels like companies feel like companies that are really going to be very careful with your data. Again, you may believe that or you may not, but Microsoft and OpenAI have a partnership for a reason, right?

00;15;56;22 - 00;15;57;02

Geoff Nielson

Right.

00;15;57;02 - 00;16;04;29

Conor Grennan

giant cloud. So I think that on the big side of the tech leader side and by the way, you do have, I mean, let's call them start. I don't know if perplexity is still a startup

00;16;04;29 - 00;16;05;25

Geoff Nielson

Yeah. Whatever

00;16;05;25 - 00;16;10;21

Conor Grennan

they're valued at now, but, you know, so these companies are worth billions of dollars and still considered startups.

00;16;10;23 - 00;16;29;20

Conor Grennan

But then on the other side, like which companies are going to win? I think the companies that are going to win are the ones that have, a couple of characteristics. And what I have seen in the past with digital transformation. And I know you've seen too, it's you're deep in this field. Digital transformation tends to work better for some companies than

00;16;29;20 - 00;16;30;21

Geoff Nielson

others. Yeah.

00;16;30;21 - 00;16;50;12

Conor Grennan

as well for, you know, industrial, you know, or like, you know, government or something like that as it would for a small startup marketing tech. Right. Because they tend to be better. But, you know, generative AI transformation doesn't really map to that at all, which is very strange when it maps to at least I've worked with a lot of people and a lot of organizations.

00;16;50;14 - 00;17;09;01

Conor Grennan

The for my own empirical research, what it maps to is if you have a leadership that is bought into this and you have a culture that's sort of a culture, you know, a like a learning culture for you organization, those are the organizations that tend to do best. Why is that? Because leadership says, look, we are going to, use AI.

00;17;09;02 - 00;17;15;11

Conor Grennan

You're going to come here, we're going to teach you AI. So I do like a whole enterprise model for, like a, you know, a digital training course

00;17;15;11 - 00;17;16;14

Geoff Nielson

and everything else. Yeah.

00;17;16;14 - 00;17;23;07

Conor Grennan

what I say to the organizations and the leaders especially, is your people want to know this stuff. They want to know it's for their next job.

00;17;23;07 - 00;17;39;11

Conor Grennan

They want to know this so they can figure out how to be a better spouse and parent and partner and gardener and cook and everything else. But also, it's just the best employee engagement strategy of all time because it's going to take their non value add tasks and condense them. So I work with a lot of big law now and nobody you know my wife's a lawyer like this.

00;17;39;17 - 00;17;43;08

Conor Grennan

There's not a lot of people that say oh I can't wait to be a lawyer when I grow up because of all

00;17;43;08 - 00;17;44;23

Geoff Nielson

the paperwork. Yeah.

00;17;44;23 - 00;17;59;14

Conor Grennan

Same thing. Right? What this does is it allows people to be more themselves. And so leaders that recognize that this is a tool to attract best talent, all that kind of stuff, all that the positive externality from that will be better productivity.

00;17;59;21 - 00;18;04;04

Conor Grennan

But it really starts with leadership recognizing that this is something that they have to get their people

00;18;04;04 - 00;18;24;07

Geoff Nielson

on board. Right? Right. That makes yeah, that makes sense. And with that, Connor, you know, I've, I've heard you say before that you're a bit of a skeptic when it comes to AI consulting. And you see AI coaching as maybe, you know, a more, appropriate or useful tactic for this, first, I mean, first of all, do you agree with that statement?

00;18;24;07 - 00;18;27;08

Geoff Nielson

And if so, can you maybe shed a little bit more light on it?

00;18;27;08 - 00;18;35;07

Conor Grennan

Yeah I think where they're I'm kind of drawing the line between consulting and coaching and look I kind of technically go down as an AI consultant to. So I'm

00;18;36;12 - 00;18;50;18

Conor Grennan

here. But I got to be honest, like the, the thing that I've seen that works and what doesn't work is that when an AI consultant will come in and just say, hey, here's this tool, here's all the things you can do with this tool.

00;18;50;20 - 00;19;10;15

Conor Grennan

And, you know, and especially like, and I work a lot with places like Microsoft and, and competitors and things like that. But when they come in, they say, hey, here's the tool, here's how you're going to use it, by the way, which has always worked in the past with Microsoft Office suites, with Google Office suites with, you know, against CRM from Salesforce and ServiceNow and everything else that tends to work.

00;19;10;15 - 00;19;28;09

Conor Grennan

It's an outstanding way of doing it, except that it doesn't work with generative AI and things like large language models. Why is that? Because with that, you really need people to come up with their own systems. You need to have people come up with their own use cases. Otherwise you're just going to give them a use case, any that they'll use it or they're not.

00;19;28;13 - 00;19;40;08

Conor Grennan

But when you get them excited and you show them again why their brain is betraying them and look and and here's the light bulb that's going to go off. And you don't want to think about it like this. You want to think about it like this, and you don't want to sort of think, oh, here's the use cases for this and this.

00;19;40;08 - 00;20;00;29

Conor Grennan

And this instead. This framework that I have is more around how do you become a better communicator and decision maker and strategize, all that sort of stuff. And the difference between consulting and coaching is consultants come in and they give you an answer to your problem. And coaches really make sure that you are, you're sort of like in a forward toward state.

00;20;00;29 - 00;20;14;21

Conor Grennan

Your brain is I'm not trying to get woo woo. I'm just sort of saying this is how it best works so that you actually lean into this and you come up with your own use cases. In other words, like a, like a personal trainer. If the personal trainer, all they did was to say, hey, look, here's how all this stuff works.

00;20;14;21 - 00;20;33;19

Conor Grennan

And vegetables work better in your body because of this. And here's how you work this Nautilus machine. That would not be very effective. Instead, you have to, inspire them, figure out like, what works best for them, and make sure it works with their schedule, for example. And it's the same with generative AI. We have to be coaching because people have to lean into this as a behavior, not just as something they're going to learn.

00;20;34;04 - 00;20;50;29

Geoff Nielson

Right. So so that role model, Connor, I'm I'm going to come back to something you said earlier, which is that for the most part, you're seeing this stay, you know, 1 to 2 steps removed from revenue, which I can imagine when you tell leaders that they might like, you know, wince a little bit because, you know, I expect they're like, but where's the money?

00;20;50;29 - 00;21;04;00

Geoff Nielson

You know, I want that. I want the money now, how are you navigating those conversations and how do you find that, you know, that's helping, organizations kind of, recalibrate their expectations here.

00;21;04;16 - 00;21;11;05

Conor Grennan

Yeah, it's it's a tough one. Right. Because organizations, when they are investing and especially these giant foundation models, I mean, the giant

00;21;11;05 - 00;21;12;03

Geoff Nielson

Yeah,

00;21;12;03 - 00;21;23;03

Conor Grennan

Everything else they're asking for huge amounts of money. They're spending billions on their own. And of course, you want companies to be like, okay, this is what's going to ultimately and fundamentally shift revenue.

00;21;23;03 - 00;21;40;22

Conor Grennan

It makes all this and a business school. It makes all the sense in the world. However, this the reason that this is a little bit different is that anything that you could just sort of take from an AI perspective and just bolt on to your product probably isn't going to give you, the best differentiation. Do you really mean

00;21;41;01 - 00;21;42;06

Geoff Nielson

Yeah. It's

00;21;42;06 - 00;21;45;09

Conor Grennan

just be able to catch up with you in six months?

00;21;45;09 - 00;21;51;22

Conor Grennan

So when you see the organizations that have really weaved AI brilliantly into their model, it's companies like Canva,

00;21;51;22 - 00;21;53;00

Geoff Nielson

right? Yeah. You know, which

00;21;53;00 - 00;22;04;26

Conor Grennan

a lot of social media people, use because now I has created what they already do and they make it better or search. Right. Perplexity is taken essentially a search engine or ChatGPT or whatever, taking a search and and giving you much better search results.

00;22;04;26 - 00;22;23;27

Conor Grennan

So people are already coming to you needing. I mean, Ikea has great, examples around this where they helped their customer service people also turn into designers. So instead of just, hey, here's where you find this type of wardrobe, it's like, well, let me tell you a little bit. And they're just using things like ChatGPT. Let me tell you about how you can use this and where this might work and how you might want to think about it.

00;22;23;27 - 00;22;45;04

Conor Grennan

Right. So they're taking something that they're already good at and using AI to augment that and to be a real leader in their space even beyond what they were before. And that will ultimately drive revenue. When I talk to companies, I say, look, you already have this workforce of whatever 1000 people or something like that, and you have those people because they are designed to work together to drive revenue.

00;22;45;04 - 00;23;07;11

Conor Grennan

Now imagine if you could increase their productivity, increase their way of thinking, and also, you know, give them new ways of thinking about the world with AI that should drive revenue. And then if they come up with a new way of using AI in their product, that's great. But the people who are going to be able to come up with those answers are the people who are already the product managers who now understand how to use generative AI.

00;23;07;11 - 00;23;13;04

Conor Grennan

And that's the thing. These ideas can come from anywhere in the organization, which is why you have to train up your entire organization.

00;23;13;17 - 00;23;44;23

Geoff Nielson

Right. Just on those ideas. I'm curious, Connor, if you have any war stories we've talked about, just the fact that, like, you know, just jumping into use cases and trying to, you know, right. AI all over them tends to not work very well because, you know, you're automating the wrong things. I'm curious if you have any war stories about, you know, AI use cases that, you know, seemed like a great idea, but like, you know, completely fell apart, or the opposite things that like, seemed like they didn't make any sense at all, but, you know, somehow got some traction.

00;23;44;25 - 00;23;45;27

Geoff Nielson

Yeah. I don't know if that's

00;23;45;27 - 00;23;51;11

Conor Grennan

if it's that they fell apart, but they were much more deceptively limiting

00;23;51;11 - 00;23;52;15

Geoff Nielson

00;23;52;15 - 00;23;54;15

Conor Grennan

than I think they thought. So,

00;23;54;15 - 00;23;55;09

Geoff Nielson

Right.

00;23;55;09 - 00;24;05;07

Conor Grennan

And I've seen this a lot, and, I mean, there's too many moments to recount, but the idea is they come in and they say, hey, for sales, here's what sales is always a great example,

00;24;05;07 - 00;24;05;17

Geoff Nielson

right?

00;24;05;19 - 00;24;05;26

Geoff Nielson

Yeah.

00;24;05;26 - 00;24;23;24

Conor Grennan

low hanging fruit kind of thing. And everybody needs sales. So in sales, what I've seen is people come in and say, hey, sales team, here's how you can use it for sales. You can use voice to do, practice calls, and you can analyze transcripts of calls and you can, you know, take entire PDFs of products and, and, you know, make bespoke proposals for organization.

00;24;23;29 - 00;24;44;15

Conor Grennan

There's a lot of ways that you can use it, obviously, but once you do those things again, they tend to get locked into those few ways of using it. If they're even using it at all. Because at that point, believe me, and you know this better than anyone, right? I mean, there's companies are getting digital products in all the time or proposals and it's like, well, our digital product will do this.

00;24;44;17 - 00;25;01;29

Conor Grennan

So when you do it that way, it just sounds like another digital product and it's like a in one ear out the other sort of thing. But what I when I talk to salespeople, the thing that I hear over and over again is salespeople want to be better. They want to be more themselves. Do I mean, they're like, we don't need another product.

00;25;02;01 - 00;25;19;27

Conor Grennan

We need more time to be ourselves. Like, give us ways that we can do what we do better. And that actually really made a lot of sense to me. So when you start cutting the shackles of salespeople and say, hey, look, you're going to spend a lot less time on this stuff and a lot more time actually on the phone and creating connections and all that kind of stuff.

00;25;20;01 - 00;25;38;17

Conor Grennan

And then you are going to, you know, almost by definition, have a client on the other side and you can talk to them about how generative AI will help them out, not because it's going to be fundamental to, you know, your product, but it just gives you a tremendous amount of credibility with your client. But ultimately, like these salespeople, they know how to make sales.

00;25;38;17 - 00;25;39;12

Conor Grennan

They know how to do all this,

00;25;39;14 - 00;25;40;00

Geoff Nielson

Yeah, but

00;25;40;00 - 00;25;54;18

Conor Grennan

they have to write things up. They have to write a proposals. And then what if you said, well, what if you could speed that up? Or what if you could clone yourself, all that kind of stuff? That to me is way better than saying, hey, here's another digital product that is going to be on the stack of other digital products.

00;25;54;20 - 00;26;00;11

Conor Grennan

Again, it's one of these things that almost starts sounding like white noise to people when they're so inundated with digital products.

00;26;00;21 - 00;26;27;10

Geoff Nielson

Yeah, that that makes sense. And it yeah, it's eye washing, right? It's just slapping slapping I on on you know, everything you can. And I think you're absolutely right I think you know consumers now can can smell that, you know, a mile away. I'm curious though Connor, we're we're at this point where it seems like almost everybody is telling us they're struggling to adopt AI at the pace that they, you know, they wish they could write.

00;26;27;10 - 00;26;49;02

Geoff Nielson

It feels like not a lot of organizations are ahead. Everybody's kind of playing catch up and yet at the same time, you've got these, as you said, these these large tech companies just pouring millions, billions of dollars into enhancing, you know, AI and trying to drive the next big thing. Do we? Where's your head out at the on that?

00;26;49;02 - 00;26;59;02

Geoff Nielson

Like, do we really need to, you know, keep pushing this forward or do we have more than enough technology here to, to keep us busy for the next 5 or 10 years? How do those two interplay.

00;26;59;12 - 00;27;19;11

Conor Grennan

More than enough technology. So, I'm. I'm so glad you brought that up. It's so funny, because, I mean, I get it, right. I get the idea that, you know, I get the idea that every company sort of like the the Google's, the Microsoft, the OpenAI's, the anthropic, the, you know, Meta's AI, etc. they all want to be kind of atop the leaderboard,

00;27;19;11 - 00;27;19;29

Geoff Nielson

Yeah.

00;27;19;29 - 00;27;34;24

Conor Grennan

And that for the best tech. And believe me, I totally get it. And it's, you know, but it actually, you know, if you think about it, if it's one of these things where if you watch a Jeep commercial or a Range Rover commercial, like what are those Jeeps and Range Rovers doing right? They are doing things like they are going over mountains.

00;27;34;24 - 00;27;37;00

Conor Grennan

These are people that are like stuck in flowing

00;27;37;00 - 00;27;37;23

Geoff Nielson

rivers. Yeah.

00;27;37;23 - 00;27;52;05

Conor Grennan

there's a hippo coming after them, and they have six people in the car and all that, like, oh my God, you've got to survive. They're doing unbelievable things. Meanwhile, I live in New Canaan, Connecticut. There's Range Rovers and jeeps all over the place. What are they driving? They're driving over paved

00;27;52;05 - 00;27;53;14

Geoff Nielson

roads. Yeah, they are parked.

00;27;53;14 - 00;28;09;26

Conor Grennan

They're driving from their home, their three acre home to the train station. So why are we all buying these things right. I have an Apple Watch on right now that can go like 100m, underwater in 18,000ft. Do you think I'm ever doing that? No, it's a it's sort of a feeling of. Oh, but it could, which

00;28;09;26 - 00;28;10;03

Geoff Nielson

means.

00;28;10;06 - 00;28;11;29

Geoff Nielson

Yeah, that. That's right.

00;28;11;29 - 00;28;30;28

Conor Grennan

Whereas what I would say is that if we just put everybody and we knew everybody was going to use ChatGPT 3.5, which was, you know, almost the original, or close to the original model that came out two years ago. And if everybody actually using it, you know, 20 times a day, we'd be much further on. So I get the tech, I appreciate the tech, and I'm all over the tech.

00;28;30;28 - 00;28;38;04

Conor Grennan

I'm posting about it all the time on LinkedIn, everything like that. But to your point, your exact point, the reality is we have not caught up with the tech.

00;28;38;04 - 00;28;55;06

Geoff Nielson

right? Right. Do you have, you know, crystal ball question when I mean, are we ever going to catch up with the tech at this point, like, or is it moving so fast that as soon as everybody's like, oh, I get ChatGPT 3.5 or 4 or, you know, four or whatever you want to call it, it's already way behind.

00;28;55;06 - 00;29;03;14

Geoff Nielson

Like is it is are we just playing catch up forever now, or is there a point where you think we can really see that, that that kind of saturation and that comfort?

00;29;03;28 - 00;29;24;00

Conor Grennan

I think that that will that moment will come when a couple things happen. First of all, people who tried this stuff early on found out that hallucinated and they couldn't trust it. And that problem is much, much lower right now, right? It's just, I mean, hallucinations are definitely still a thing, but the problems are. It's just not a huge problem anymore.

00;29;24;00 - 00;29;41;29

Conor Grennan

Right? It's it's especially now that you have ChatGPT connect to the internet, Gemini connect the internet, etc., etc.. So that's not a huge issue. So I think that if we if we knew that we solved the hallucination problem, that could be somewhat helpful. But I think the much bigger, sort of thing that will get people really going is user interface.

00;29;41;29 - 00;30;04;00

Conor Grennan

I talked about this all the time. The problem with something like ChatGPT is not that it's not unbelievably powerful. The problem is that it doesn't look like, Jeff Bezos is talking to you. So if you have a marketing question, you can get responses on maybe on the level of Jeff Bezos, but really brilliant marketing feedback, right? The problem is it looks like a Google search bar.

00;30;04;07 - 00;30;19;07

Conor Grennan

And so when your brain seizes because your brain, automates things, and when your brain sees something, it automatically sort of like. So if you see a baby, you automatically know how to talk to that baby like a baby or a puppy or something like that, because you don't have to think about it in the same way when your brain sees a kind of a chatbot like that, it's it's like, I know what this is.

00;30;19;07 - 00;30;38;10

Conor Grennan

It's like a Google search command response. This is not this is sort of a more subconscious thing, but I think more and more they're starting to crack various slowly the user interface problem, which is when you're using this, how do you get people to understand that you're talking to this like a teammate, like a new colleague, and not like a search bar?

00;30;38;10 - 00;30;42;15

Conor Grennan

That's really challenging. But I think that that's where the real movement will start taking place.

00;30;42;15 - 00;30;56;12

Geoff Nielson

And what does that look like in your mind, Connor? Like, is that like, do you put a Wizard of Oz head, like in the screen or Gandalf or like. Like what? You know, just a regular person. What what's the interface do you think that can can drive better adoption or like, you know, better understanding

00;30;56;16 - 00;31;16;21

Conor Grennan

Yeah, it is such a mystery. I'm not sure. I mean, I'll say what I've seen sometimes, which is Google. I've gotten to work a bunch with Google, and I've seen some of the things they're kind of doing behind the scenes a little bit with certain things. But like, if you had a chat bot, right, that was, you know, the chat bot was, the how to get advice from Sam Altman on how to, you know, create a new startup.

00;31;16;24 - 00;31;31;28

Conor Grennan

And it actually looked like and by the way, they do incredible video avatars. But if it was actually sitting there on your screen and looked and sounded like Sam Altman and you're like, hey, Sam, how do I do this? And Sam, I'm just as I'm talking right now because they get are they're getting very good. They're like, okay, so here's how I would probably do it.

00;31;32;00 - 00;31;53;21

Conor Grennan

It's a very simple kind of thing, but I think that that would actually go a long way. Now. I think that there's a lot of problems with that. I'm not sure that they can actually pull that off, but that is the sort of direction. Because again, think about it like if you were to, I use this analogy sometimes if you were planning a trip to Costa Rica and here you had a laptop with a, you know, top ten things to do with my family in Costa Rica.

00;31;53;21 - 00;32;08;27

Conor Grennan

But at a desk right next to it was the head of the Costa Rican tourism board. You would talk to the head of the Costa Rican tourism Board in a much different way then you would, you know, like ChatGPT or something like that, right? Even though they can offer pretty much the same advice, but just because it's a human.

00;32;08;27 - 00;32;13;16

Conor Grennan

So I think that user interface, I'm not sure how we crack it, but I think that that's going to be the next big frontier.

00;32;13;16 - 00;32;14;11

Geoff Nielson

right?

00;32;14;13 - 00;32;41;00

Geoff Nielson

Right. And it seems like, yeah, it just unlocks a whole world. More possibilities if people are looking at this right now and they just see a Google search bar, right, versus seeing, you know, a much broader tool than that. Just on the tool side, you know, ChatGPT obviously, you know, when it came on to the scene, you know, watershed moment, you know, for technology, do you see another watershed moment for AI, for any other technology coming in the next few years?

00;32;41;00 - 00;32;47;17

Geoff Nielson

Or, again, is this like, you know, we've got our hands full for the next, you know, five, ten years with this?

00;32;47;17 - 00;33;08;20

Conor Grennan

I think that, you know, they talk about or we talk about AI agents a lot. And the challenge with AI agents right now is that they are very ill defined. So Microsoft sort of says they have AI agents. They don't they don't really have AI agents built into copilot. They have something they're essentially what we call custom GPT, which aren't really AI agents.

00;33;08;22 - 00;33;25;00

Conor Grennan

Salesforce is talking a lot about AI agents. Anthropic, which has the chat bot, Claude has something called computers, which is an AI agent. And kind of you just say, hey, I need to plan a trip, and it kind of takes over your computer and it starts just doing the work for you. AI agents will definitely be the next thing.

00;33;25;00 - 00;33;52;22

Conor Grennan

And when we are talking about really driving revenue, they are going to be huge because I mean, you know, Jensen Huang of Nvidia talks about having a company of 100 million agents. What does that mean? Well, it just means that these are little tools that can do. And it would have to be probably a fairly repetitive task, but to do it reliably and to do it, you know, if you just sort of say like, hey, I need to have this thing done and you instead of turning to a person, you turn into a turn to an AI agent that actually goes and doesn't just give you information back, but it actually goes and accomplishes the

00;33;52;22 - 00;34;09;19

Conor Grennan

task. That's really huge. I mean, I think kind of chat bots almost for customer service are almost like this, but they're not they're not quite because they're not doing complex tasks. But I think that that is the next tech that will actually really start to move revenue. But the problem is we still have to understand how to train them.

00;34;09;21 - 00;34;22;29

Conor Grennan

And so I think there are some companies that are waiting for that. But again, everybody who knows these these sort of tools and this technology says you still have to train an agent. It's still acting like a human. You have to sort of give it feedback and iterate and see what good looks like. Give it examples, etc., etc..

00;34;23;02 - 00;34;29;08

Conor Grennan

So once we start hitting that, we still need people who understand large language models who can train those new agents.

00;34;29;08 - 00;34;41;24

Geoff Nielson

right? Right. And once we get, you know, if we're coming to this age of AI agents, there's, you know, and we see this so often right where it's like kind of an arms race for these agents. And I think, you know,

00;34;41;24 - 00;34;45;03

Geoff Nielson

by the way, it sounds like that's already begun. You know, do we get to this

00;34;45;03 - 00;34;54;26

Geoff Nielson

point of proliferation where like, you know, every website you go to or every organization you interact with has its own, you know, independent, siloed AI agent?

00;34;54;28 - 00;35;11;04

Geoff Nielson

And, you know, do we end up in that world to someone, you know, like a Google or an Apple, create an agent of agents, you know, and any, any, you know, kind of forecasts for, for, you know, where this will take us in the next few years.

00;35;11;13 - 00;35;34;10

Conor Grennan

Yeah. I think I don't know the timeline for it, because it's getting so strange with the way they're talking about it too. Right. So AGI, as defined by Sam Altman at OpenAI level five of AGI or sorry, level five of sort of like the progression and evolution of generative AI is what they call AGI artificial general intelligence. And that's when a, an agent should be able to command all these other agents and be able to run a company.

00;35;34;12 - 00;35;51;16

Conor Grennan

Now, I think they're backtracking on that a little bit. They're sort of calling that superintelligence. So I don't know if that's a year away or something like that. But yeah, absolutely. I mean, and I can imagine how it would look like if I want to book a trip through, you know, kayak or something like that. It's just going to be, well, tell me, like what you want to do.

00;35;51;16 - 00;36;07;11

Conor Grennan

It's like, well, I want to go visit my, grandparents up in Minnesota. It's like, all right, so let's talk about this. Like, do you probably need a car? What about this one? And and it can do it all super fast. I can imagine what that would look like. I just don't know how real I think they're going to be very unreliable early on.

00;36;07;11 - 00;36;31;18

Conor Grennan

And that's going to, again, probably screw up, our comfort with using these things. So I think that even if these things came overnight, I still think that unless they came out of the box perfectly, which would never happen, we're still going to have that challenge, because when they're not going to work very well and people are going to get disenchanted, as they did with hallucinations on these early models and things like that, I still think it's going to be pushed out further than people think.

00;36;31;18 - 00;36;44;21

Conor Grennan

I don't think it's a matter of switching on the technology and people use it because believe me, there's a lot of tech like I use clawed all the time. Anthropic clock I often can't find a lot of people if I'm just going out to a company that even know what Claude is, and meanwhile I can't live without it.

00;36;44;21 - 00;37;03;15

Conor Grennan

It's it's the best writer of any of the large language models by a wide margin. And then a lot of tools on ChatGPT, like a round like code interpreter, which interprets spreadsheets and voice and vision. These are not things that people tend to use, and they've been out for a year. And so I just think that behavior is going to be very slow, trying to catch up with the technology.

00;37;03;15 - 00;37;19;13

Geoff Nielson

it's almost a cliche at this point to say that we're at, you know, kind of, generational, you know, technology change with I here. I'm curious, just like scale of 1 to 10, how exciting is the period we're in right now? And what excites you most about the potential and the possibility?

00;37;20;16 - 00;37;40;19

Conor Grennan

So I'm a big, like, give everything five stars thing. So you won't be surprised by my ten. But, But I just think that we are in a period where it's almost unlike anything we've seen. I think the internet was similar to this, but, but even with the internet, it's, you know, there was a slow ramp up to understanding how we could use it.

00;37;40;19 - 00;38;00;04

Conor Grennan

I mean, it started off with just, like, reading the news, and then there was email, but then now people are completely online. I mean, when you look at Walmart and everything else, these are an Amazon. My gosh, Amazon would not literally not exist without the internet and is the most powerful, buying platform on the planet. And it's taking over absolutely everything.

00;38;00;04 - 00;38;26;28

Conor Grennan

And so and we're not that far removed from when Amazon really came out, when people started to think about it as something they could use every day. So when I think about something that when we think about, well, what does ChatGPT sort of replace or what are the large language models replace? And when you hear answers like, well, it kind of replaces like the human and the brain, that to me is a different level of existential opportunity and threat than anything that like a new buying platform.

00;38;26;28 - 00;38;53;17

Conor Grennan

So I think that anybody that's feeling like, oh, this is overhyped is really not seeing what this can be and what it is. And the and the fact that we're kind of basically two years in and where we have gotten to after two years. So I would say that this is going to disrupt everything. And the interesting part about that to me is when organizations are trying to do things the way they've always done them, even, for example, planning, you know, 1 to 3 year strategic plans, how do you even do that?

00;38;53;17 - 00;39;05;10

Conor Grennan

I don't know what the world is going to look like even in two years. Whereas in the past, you always sort of had a general sense. Well, you know, we'll have to, you know, track along with how the economy functions. And, but nobody's ever thought, well, what if what if aliens came in a year and a half, right?

00;39;05;10 - 00;39;22;12

Conor Grennan

Then things would probably be different. And I feel like we're in the mode of what if aliens arrived? Because if this thing really can totally disrupt how your workforce does everything and controls, you know, revenue and people management and everything else, that's the level of, of opportunity that we're talking about right now.

00;39;22;12 - 00;39;24;17

Geoff Nielson

that that makes complete sense.

00;39;24;17 - 00;39;29;28

Geoff Nielson

You got my you got my gears turning corner in a couple of different ways. There a couple of questions I want to ask as a follow up. The first one

00;39;29;28 - 00;39;47;08

Geoff Nielson

is, you know, you compare it to the, you know, the advent of the internet, which I completely buy, by the way. One of the things that happened, though, with the advent of the internet, as you know, you probably remember, is like, you know, eventually it did take off and everyone, you know, kind of jumped on e-commerce.

00;39;47;08 - 00;39;52;11

Geoff Nielson

It transformed, you know, everything. But along the way, we had this funny thing called the dotcom

00;39;52;11 - 00;40;10;05

Geoff Nielson

bubble, right? Where, where like, you know, everybody, you know, put, you know, web or internet before their company name. And, you know, there were winners or losers that we couldn't predict in those first few years. Are we in a similar bubble now with AI, or have we learned the lessons from, you know, last time around?

00;40;10;07 - 00;40;10;25

Geoff Nielson

Oh my gosh,

00;40;10;25 - 00;40;27;05

Conor Grennan

I'd love if we learned lessons. I think that there's two types of bubbles that we're looking at right now, and I think we may have already burst the first bubble. That first bubble was, all the wrappers on top of this stuff, right? I mean, even if you go now, I can't tell you how many people say, like, okay, can you tell me download ChatGPT.

00;40;27;06 - 00;40;45;16

Conor Grennan

And I downloaded it, but it doesn't work. I'm like, right, because that app came out way before the ChatGPT app and still has a billion downloads or something like that, or all these things where it's like, oh, we built ChatGPT into Excel, and now you can use it in your spreadsheet. The problem is OpenAI, Google, all these other ones are just steamrolling those companies.

00;40;45;18 - 00;41;13;01

Conor Grennan

And that's the the cheese graters.com moment, right. Where like anybody that put.com after a thing could, you know, get tons of you know, venture capital or something like that. So I feel like we actually may have spent the first 12 months blowing through that part of the.com bubble. I may be wrong on that, but I don't think anybody is like looking to run out and throw themselves in front of a tank like Tiananmen Square style and say like, don't worry, I can.

00;41;13;01 - 00;41;35;13

Conor Grennan

I can handle this. Right? It's like those there's there's probably 5 or 6 huge companies that no matter what can you can do they can do better. It's just a fact. And that's a little disappointing. Not that other services can't be offered, but in terms of trying to compete with them would be very, very expensive. The other valuation thing, which I don't really know, is, well, what about sort of, you know, top tier and then second tier, right?

00;41;35;13 - 00;41;50;25

Conor Grennan

I mean, there's what I'm seeing now is Ilias Tusk of our leaves, OpenAI, one of the founders, and immediately gets $1 billion for, safe, safe superintelligence or AGI or whatever it was called, raises $1 billion saying we're not even going to have anything for two years. Like what is

00;41;50;25 - 00;41;51;12

Geoff Nielson

Yeah,

00;41;51;12 - 00;41;56;29

Conor Grennan

Same with the mere morality of the world leaving anybody that leaves a big company, they immediately get $1 billion.

00;41;56;29 - 00;42;12;19

Conor Grennan

And that just feels like, these investors hedging their bets. It's like, well, we don't know where it's going to go, but we know it's going to go somewhere around here. We don't want to miss the boat. So I am I do think there's a bubble there in terms of the investment, I, I don't think there's a bubble.

00;42;12;22 - 00;42;21;19

Conor Grennan

If, you know, Bloomberg or JP Morgan or somebody that like invest in this, I think that that is a wise thing. Now, there may be a bubble in building your own tool that

00;42;21;19 - 00;42;22;04

Geoff Nielson

Right.

00;42;22;04 - 00;42;29;03

Conor Grennan

I don't usually advise that unless you're really huge, like a McKinsey or something like that. But I do think that we've probably burst the first bubble.

00;42;29;03 - 00;42;42;22

Conor Grennan

I may be wrong, I might be wrong, but I don't think anybody's full interest to jump in front of, the open air steamroller anymore. But the second thing about, how these, you know, the perplexity, valuation, all these valuations, are they overblown? I don't know, maybe.

00;42;42;22 - 00;42;46;09

Geoff Nielson

Right, right. So it sounds like, you know, you're bullish on the technology.

00;42;46;15 - 00;42;57;27

Geoff Nielson

You're bullish on, you know, individuals and broadly organizations taking advantage of this. But when it comes to the valuations of individual organizations there's still going to be winners and losers. Is that fair

00;42;58;01 - 00;43;08;28

Conor Grennan

I think so. I mean, like, what are we doing with Nvidia. You know Nvidia is I how many. You know it's like it's worth trillions of dollars. Now Amazon comes out and says hey we have these chips. And guess what. Apple's using these chips. Anthropic is going to

00;43;08;28 - 00;43;09;09

Geoff Nielson

I'm

00;43;09;09 - 00;43;12;20

Conor Grennan

chips. Like what does that do to Nvidia. Was that overblown.

00;43;12;20 - 00;43;13;03

Conor Grennan

Maybe

00;43;13;03 - 00;43;13;18

Geoff Nielson

Yeah.

00;43;13;18 - 00;43;18;14

Conor Grennan

can you know trounce their fourth quarter projections and still their stock drops because

00;43;18;14 - 00;43;18;26

Geoff Nielson

yeah

00;43;18;26 - 00;43;38;09

Conor Grennan

places like Amazon other everybody else is building these things out cheaper. So yes Nvidia may have the you know it's sort of like the Rolls Royce of this kind of thing. But do people need the Rolls-Royce? That's the thing. Like all these language models, they can start being built open source everything else on phones, on devices like do we need the absolute top of the line stuff?

00;43;38;09 - 00;43;44;03

Conor Grennan

I don't know. So in those cases, I do think probably they're overblown. But again, I'm not sure I might be wrong.

00;43;44;03 - 00;43;49;25

Geoff Nielson

Yeah, yeah I do I do want to come back to, you know, the transformative potential of this technology.

00;43;49;26 - 00;44;10;27

Geoff Nielson

And you talk about, you know, this, you know, there's this narrative about it replacing people and kind of transforming organizations. Are we going to get to the era of like, you know, the autonomous enterprise, the AI CEO, if you can call it that or, you know, is is it much more just, you know, being able to talk to an avatar of Sam Altman who can give you advice?

00;44;11;08 - 00;44;28;19

Conor Grennan

Listen, I tend to believe the people who are creating this stuff and everything I'm hearing from Zimmerman, who has OpenAI, Dario Amodei, who has anthropic, Sundar Pichai, who has Google in that says Gemini, etc., etc. they're all talking about that not just in terms of someday, but like in the next few years.

00;44;28;19 - 00;44;29;10

Geoff Nielson

Yeah, and

00;44;29;10 - 00;44;36;10

Conor Grennan

I think they would have a lot to lose by start, you know, putting this stuff out and this stuff just falling apart, which is interesting, right?

00;44;36;10 - 00;45;00;08

Conor Grennan

Because you have you're the younger Koons, who's my colleague over at NYU, and is obviously runs, you know, Meta's AI. So obviously a huge, hugely bullish everything else. But what he talks about is that this thing has the intelligence of a house cat. So how you get from a house cat to running an organization in 18 months or two years as the head of anthropic head of OpenAI are talking about I don't know how you square that because I think brilliant people can disagree on this.

00;45;00;10 - 00;45;11;12

Conor Grennan

I don't know about running an organization and and a CEO. I think that it's probably, again, sort of like an Iron Man suit. If you compare Tony Stark and what he can do versus what he could do in an Iron Man suit,

00;45;11;12 - 00;45;12;00

Geoff Nielson

Yeah,

00;45;12;00 - 00;45;20;23

Conor Grennan

things are wildly, wildly different. But I still think that you need I don't even want to call it the human in the loop, but just the human understanding what risk is and where to guide it.

00;45;20;23 - 00;45;25;25

Conor Grennan

And who are we and everything else. It's hard for me to imagine a completely autonomous

00;45;25;25 - 00;45;26;10

Geoff Nielson

right

00;45;26;10 - 00;45;30;08

Conor Grennan

for many, many years, but I don't know that we're not heading that direction.

00;45;30;08 - 00;45;41;17

Geoff Nielson

Yeah. So you still see it though, is you need the Tony Stark and the Iron Man suit can't run. You know, the stark. What is it called? Stark corporate. Whatever. Stark enterprises just yet. Yeah yeah, yeah,

00;45;41;17 - 00;45;54;25

Conor Grennan

yeah, I think so, I think so, I mean, again, you know, to take that analogy to get Marvel geeky, I mean, like that iron Man, you still can go off and do stuff, but it has to be essentially guided by a human. I think that these agents, everything else, will be able to do very narrow things.

00;45;54;25 - 00;46;04;12

Conor Grennan

I mean, you know, when you talk about jobs and things like that, you're talking about, you know, if if a Sam Altman said something a long time ago that I really liked, which was this thing doesn't really take jobs, it takes

00;46;04;12 - 00;46;05;18

Geoff Nielson

test, right

00;46;05;18 - 00;46;16;20

Conor Grennan

job is made up only of a few tasks and let's call it customer service, let's call it paralegal or things like that or something where it's, you know, your task is to summarize documents and give people the relevant thing.

00;46;16;22 - 00;46;36;15

Conor Grennan

Large language models do that very well and so on. Or, you know, image generation or graphic design, things like that. It's not that you don't need graphic designers. I think that's actually really a critical point when I, you know, I have my friend that I work with, Robert Haslam is, is a brilliant designer and good with AI, but it's not just now that I have, I can replicate what he does.

00;46;36;15 - 00;46;38;22

Conor Grennan

Now. He does all this stuff brilliantly,

00;46;38;22 - 00;46;39;10

Geoff Nielson

Yeah.

00;46;39;10 - 00;46;54;08

Conor Grennan

strategist because he understands esthetic what things should look like and understands how to guide tools like Midjourney and Dall-E and things like that. So you can't just replace those people, but you may need fewer of them. If a lot of this, what they do is just labor.

00;46;54;08 - 00;46;56;07

Geoff Nielson

So I'm gonna I'm going to use that as a jump off.

00;46;56;07 - 00;47;17;08

Geoff Nielson

Connor on you know, something I've heard you say before, which is that, you know, you think that prompt engineering is kind of B.S. if I can, you know, put put those words in your mouth. How does that, you know, how does that intersect with the comment you just made? You know, is it a separate role? Is it just making everybody smarter or everybody able to talk the language of, you know, AI of these tools?

00;47;17;08 - 00;47;18;27

Geoff Nielson

What does that look like?

00;47;18;27 - 00;47;38;05

Conor Grennan

Yeah, I think that's I think that's fair because you're right. There's there there may be a little bit of, I'm trying to think of how how to connect these two, but here's how I would think about it. If you think about prompt engineering, there's a lot of just courses out there like here's how to prompt engineer. But in fact, there's not any special thing you have to tell ChatGPT that you wouldn't tell a colleague.

00;47;38;08 - 00;47;54;27

Conor Grennan

So if you look at anything like, let's say you had a new colleague come on board, what would you do with that person? Well, you'd give them. So I have a digital module built out around this, right. Like you'd give them a task. Right. And then and you'd sort of also say, and here's the other additional information you probably need, like I'm going to ask you to write this thing.

00;47;54;27 - 00;48;09;19

Conor Grennan

So here's how these things have been written in the past. And here's our statement of values. And here's the thing that you're writing about. Here's some extra information about right. You give them some additional information about that, and then you set them off to write their task. And then when they come back, chances are they haven't done it perfectly.

00;48;09;23 - 00;48;15;18

Conor Grennan

So instead it's like, oh, well, you know, that's good, but you want to and then you start to refine it with them, right? You kind of go back

00;48;15;18 - 00;48;16;13

Geoff Nielson

to. Yeah,

00;48;16;13 - 00;48;22;12

Conor Grennan

that process I just described could be used exactly for a new colleague or

00;48;22;12 - 00;48;23;15

Geoff Nielson

Yeah.

00;48;23;15 - 00;48;27;28

Conor Grennan

with no edits, right? No edits. It's exactly that process.

00;48;27;28 - 00;48;40;09

Conor Grennan

You give it a task, you give it all the context, you give it examples of what looks good, and then you refine it. That's it. So when people say, oh, you need to learn prompt engineering, I'm like, why like this? All you have to do is talk to it like the same way you talk to a human and you already know how to do that.

00;48;40;11 - 00;48;57;27

Conor Grennan

So there's nothing to learn in that regard. However, the way I think it connects with what we were just talking about in terms of, well, what about people who are really good at that? I guess what I would say is it's the same thing, like if you have a graphic designer and you're trying to do a midjourney, like one of these AI image generation,

00;48;57;27 - 00;48;58;22

Geoff Nielson

Yeah,

00;48;58;22 - 00;49;05;08

Conor Grennan

like Dall-E from ChatGPT or Midjourney or something like that, they still would know how to do it.

00;49;05;08 - 00;49;24;06

Conor Grennan

Right. So when I see my friend Robert do this stuff, he has like an esthetic that he's describing, you know, to this he's, he's putting in like the, you know, hex colors or whatever he's putting like it should fade to this, it should feel like bubble because that's how he would do it if he was just designing it with, you know, pre AI and everything like that.

00;49;24;06 - 00;49;27;09

Conor Grennan

Whereas I'm just saying like, hey, create a beautiful, like magical landscape

00;49;27;09 - 00;49;28;11

Geoff Nielson

right? Or

00;49;28;11 - 00;49;37;13

Conor Grennan

or something like that. I just I wouldn't know. And so I think that even the people who do it, all they're doing is stating in the same way they've always stated it, it's just into a new technology.

00;49;37;13 - 00;49;41;23

Geoff Nielson

Right? The knowledge is already in their brain. It's just communicating it right.

00;49;41;23 - 00;49;46;22

Geoff Nielson

Thanks again Connor really appreciate it. This has been super fun. Thank you so much for having me. I really would be