00:00AI is a different growth story, but it is changing some of the math about how we think about investment
00:05returns inside organizations, but it is still very, very early days. The majority of organizations
00:11that are bringing AI in are not really seeing growth gains. They're seeing a bit of productivity
00:16or cost reduction, but we're not really seeing a lot of the use case that are generating
00:20detrimental revenue or value creation for organizations. Again, we're early days, but
00:24the last couple of months are showing a little bit more doubt about how much these investments
00:28are going to translate into returns. So when you talk to your C-suite clients,
00:32are they looking to embrace AI? Are they still in the stage of what is AI? And is it a
00:37friend to me?
00:38How do I think about it? What's the conversation you're having? So I love this question because
00:41there's been a transition. So we went from kind of 2023, I'd say it was doubting. CEOs did not want
00:47to jump on another trend that was not going to pan out. Then we went into dabbling, right? Use cases,
00:53throwing co-pilot licenses at people. We are very slowly transitioning into what I call deliberate,
00:58actually bringing AI into strategy. But relatively early days, the majority of companies are still
01:03dabbling. So while AI adoption is very high, it is incredibly shallow. You know, only kind of 10 to 15
01:08%
01:09of business work is being done in any form with AI. And any kind of cross-team, cross-functional use
01:13cases are relatively small.
01:15What is the, as we step back here and think about some of the, all the economic data we've
01:19been getting over the last several weeks, including this week, we'll get some more
01:22on the labor front here. What's the conversations you're having today with your clients about the
01:27economy? What are you hearing? What's the conversation?
01:29You know, what's funny is that the market gets bored of topics when it goes on too long. You know,
01:34we're kind of a bored of the Iran war, we're a bored of the K-shaped economy. CEOs don't have
01:37the
01:37luxury of getting bored. Just because a trend has been going on for a couple of months doesn't mean
01:42that it affects strategy. So I'd say, look, we're looking at the Iran war. We are looking at the
01:46economy, especially the lower end, and tariffs are back in play as of yesterday. And again, we might
01:51be tired of talking about tariffs, but there are real economic income companies trying to make some
01:55of these decisions.
01:56The Pennsylvania Railroad was religion in my family, arcing across to Chicago. I mean, it just,
02:02it's what came out of the three debacles of railroads going back to 1890. The analog to the
02:09railroads and AI, where are we in that path right now?
02:13Very early days. The difference is, 1890-ish. The difference is, Tom, when we laid some of the
02:19early railroads or the kind of the cable infrastructure in the 90s, there was nothing
02:23really going on them or through them in the early days. We do not have a demand issue right now.
02:29Like
02:29this is very much, we are still lower supply than we have demand. So demand is high, it is increasing
02:33at a nonlinear rate, and it will continue to. But we are going to see a very volatile market when
02:38it
02:38comes to AI. We're seeing a shock the last couple of weeks, because a lot of the vendors are changing
02:43how they're pricing, which is why compute cost and token max and all of those things are coming
02:47into the conversation.
02:48Are the bro listening to people like you? Do you feel like they're in a vacuum in their
02:52earnestness and their $8 lattes out on the West Coast? Or are they actually listening to
02:58Bryn Jolson or Hamas about what's actually going on?
03:01You know, if you look at the differences between the impact that both economists, academics,
03:07and tech firms think that AI will have on productivity, the range is between 0.1% and
03:1180%. Like that is a hundreds of order magnitude difference. Like we will eventually need to
03:16ladder out somewhere. It'll be higher than we've seen in other tech booms. But at this
03:20point, I think we're dramatically overestimating the near to short term impact that AI will
03:24have on productivity.
03:26Because I mean, that's probably that's a non consensus view, I would argue, because I think
03:30most people that come into this studio with Tom and I suggest that, I mean, you got electricity
03:35at the top of the compendium. And right below that is AI. Well below that is the internet.
03:41That's kind of where people kind of weigh this because adoption was so much higher. You know,
03:45we reached 50% of global adoption in three years, the internet was seven to 10. The difference is
03:50what is happening inside organizations, it is very, I don't want to say easy, because nothing's
03:54easy. But it is very approachable to get productivity booms at a team or function level. So engineering
03:59team at scale 30% productivity, customer service support 30 to 50% productivity. What we do not have
04:05is cross team cross functional organizational wide productivity, which is something you will see
04:09as an external. And the reason is getting those kind of gains does not take throwing AI on top of
04:15work. It takes actually redesigning workflows. And that is really hard work that takes time.
04:20So we'll get there maybe slash probably, but short to near term, we are looking at change.
04:25Paul's too polite. So I'm going to ask, it's just another corporate education program.
04:31Baloney, it's Fuqua, the future of leadership now. What's the distinction about Duke corporate
04:37education versus another soiree to look buff and beautiful? I think all forms of corporate
04:43education, whether it's London Business School or Duke, we are trying to help deliver impact and value
04:48creation and really understanding not just the topics, but how they can practically apply it inside
04:52their day to day work. Do you make them do math? All the time. This is cool. I get I
04:57get I get
04:58up charts. I always caveat it's going to be a chart. I need Duke corporate education to sponsor
05:03me. My first sponsor, the first people that said the idiot with the bow tie, let's throw money at
05:08them. Stern School of Business NYU. And one of the students showed me their math. Paul. Yep.
05:15I was like, are you kidding me? That's the energy that's in these programs. Math still matters,
05:20Tom. Math still matters. She can come back. Exactly. Exactly. How do you put it? How do you
05:26put AI into the curriculum here? Because I would think that would be great. If I'm anybody north of
05:3228, 29 years old, I need to be educated on AI. So there is plenty of education. I look at
05:38the
05:38vendor front as well on what we call hands on keyboard education. And that's fine. But that's not what
05:44leadership teams need. Leadership teams need to actually understand how do I retrain my teams? How do I
05:49hold people accountable for AI? And not to kind of overwhelm on jargon, but most companies are great
05:54at tracking the outputs of AI, how many use cases are in play, how many people have licenses, how many
05:59tokens are used. Organizations get nothing from those output measures. What we really need is kind
06:05of this organizational-wide, leadership-wide training on how to deliver outcomes from AI versus
06:09just outputs. How about the ethics of AI? I mean, I was speaking to the Boston Consulting Group
06:14recently, and they say before they engage with a client who wants to, quote unquote, implement AI into
06:19their business model, they want to get buy-off at the C-suite level and the board level on
06:24ethical implementation of AI. How does that figure into it?
06:27It absolutely does. And for full disclosure, I'm an advisor at BCG as well, is that if you think
06:31about the, first of all, it's very much, it's not a senior leadership thing as well. Younger
06:35employees are very, very concerned about the ethical use as well. If you pull younger, excuse me,
06:39pull younger employees, that's what they want to talk about as well. The challenge is there's three or
06:44four simultaneous planks you need to work on if you want to get benefits from AI. You need to do
06:50organizational-wide upskilling. You need to focus on the governance, ethics, responsible AI. You need
06:54to build your tech infrastructure data platform. Most established organizations don't have data
06:59in a way that they can actually leverage it using these AI tools, and then you've got to link it
07:04to
07:04outcomes. So doing those four things at the same time, we call it parallel pathing, is very difficult.
07:09But if you don't have the governance, responsible AI in place, you're not going to move. The irony is
07:13that boundaries give folks freedom. So organizations that are moving the fastest are those that have
07:19the clearest boundaries in governance and responsible AI. When the rules are clear, people move faster.
Comments