Conor Grennan has become the go-to expert for bringing the benefits of large language models (LLMs) to businesses. With a friendly and human-first approach, he advises teams from a range of industries — including ServiceNow, PwC, Citadel, and even the U.S. government — to help them embrace LLM technology in their daily operations.

It’s the latest endeavor for this New York Times best-selling author, whose insatiable curiosity and fascination with the brain has led him along an eclectic career path culminating in his recent appointment as Chief AI Architect for NYU Stern School of Business. We were lucky enough to catch up with Conor to hear about how he’s helping to shape the way companies think about training employees to use generative AI. Check out what he had to say below.

0:00

/3:44

Conor Grennan on how to use GenAI

Conor’s journey from writer to AI coach may be less of a random walk than it first appears. “English majors are gonna rule the future of AI,” Conor recently stated on TikTok, one of several social channels where he has amassed tens of thousands of followers. Labor economists and education experts increasingly agree. They cite critical thinking and language skills as the best foundations for interacting with the new generation of AI that relies on natural language dialogue rather than coding skills.

This perspective serves as a major cornerstone of the training that Conor delivers, along with the persistent finding that successfully incorporating new technologies into the enterprise hinges more on changing hearts and minds than technical challenges. Conor focuses on the behaviors and mindset that enable users to make the most of LLM-based tools, as opposed to teaching step-by-step basics or specific prompt recipes.

The insights he’s gleaned through his work with companies and users, such as the way the brain’s autopilot mechanism stops us from “talking” to chatbots like humans, present helpful considerations for developers and executives alike.

An edited version of our conversation follows.

Tell us how you went from author to AI expert.

After living abroad for many years, and doing a lot of public policy work relating to trafficked children on the other side of the world, I came back to the U.S. I wanted to take the next leap, so I got into the MBA program at Stern to pursue finance. But then I got a book contract along the way and ended up becoming a writer. And then, about four years after that, I went back to Stern and took on the role of Dean of MBA students there, which was a great ride.

A month or two after OpenAI released ChatGPT, I started looking at it and pretty quickly thought, “Holy cow, this is going to change everything about everything.” Also, because I have kids, I thought that this is going to be a crazy plagiarism machine that we're going to have to tackle, so I better learn how to use it. But also, importantly, I was thinking about how our MBA students were going off into this world where this would be really transformational. I was going to need to build out a framework to teach them, as well as the faculty, how to use it.

To figure that out, I ended up talking to heads of companies, usually in financial services, asking, “Hey, how are you all using this?” And the answer that came back in early 2023 was: “Well, we're not really using these tools.” I found myself telling them how powerful it was and things sort of flipped — I found myself training their C-suite on generative AI.

It's strange to go from the world of academia into training on generative AI. I think it would have otherwise been impossible if it wasn't for the very strange sorcery that is generative AI.

What were some of your early learnings when you started working with generative AI tools?

One of the biggest things was that there seemed to be no learning required. It was a very flat learning curve. And that was strange because it's unusual to not have to find a user manual or watch videos or take a class for software that's this powerful. In fact, what I started to realize was that the only thing I seemed to really need to know how to do was to talk to it like a human.

But our brain is not used to talking to a laptop like a human. It would be like saying, “Talk to this baby like a college professor all day.” It's very simple in principle, but try to do it, and you can't because your brain automates things like that. It’s fascinating, and it took me down this path of realizing that this wasn't really about teaching a tool — it was about teaching behavior when using the tool.

How do you teach the right behavior?

Executives often ask me, “What's the first step? What's the second step?” I give them a totally different way of looking at it. Forget about this as a tool. Forget about use cases. Let's talk about what's actually happening in our brain when we are interacting with a large language model. And then I take them through a framework that is completely industry-agnostic, department-agnostic, and role-agnostic.

I used to go into a healthcare company and they would say, “Give us the top 10 uses for healthcare” and I would provide them. The problem was that when I came back a little while later, they were not doing them or still doing only those same use cases. So, what I say now is “Let's go through this and then you're going to be telling me all the use cases for healthcare.”

The analogy I use sometimes is that if you could go back in time and give people a light bulb, they would say, “This is incredible. This is gonna change everything.” But if you went back in time and gave them electricity, they’d say, “Well, this is great, but how do I use this?” That's a very hard question to answer because it's so broad.

And electricity doesn't allow you to do new things necessarily, right? It allows you to do things better and faster. And that's really powerful. Generative AI is the same — it's not doing something that you could never have done by yourself. You can come up with a new performance management system or new marketing tools. But LLMs allow you to do those things faster and at higher quality.

When you start showing that, that's where you really start to make a strategic impact on a company.

What are some of the misconceptions you encounter when working with companies?

The number one misconception about large language models is the security part. I’m not saying we should all start uploading proprietary data. But I think people have a different sense of what “proprietary” means. They think everything's proprietary when it's really not.

The other thing is that there's a sense that if you've uploaded something like a client list for a big bank, it's akin to leaving it on the table in Starbucks. That's just not how these tools work.

Enterprise-grade systems are important and easily available if you have to protect your data. So, security shouldn’t be a stumbling block if you put good guard rails in place, and if you trust your people.

Do you have some examples of companies that you think are getting the employee upskilling right?

The honest truth is that very few are doing it well. And when I say doing it well, I mean that it's really infused across the workforce.That's not a knock or anything like that. It's just that this is still a somewhat misunderstood technology.

The companies that are best positioned to do it well tend to be two different types of organizations. One is a fairly small organization where everybody in the building has three jobs and they're just looking for anything that will help them. The other is those with strong cultures, especially with a strong C-suite that's built a lot of trust and a great authentic learning culture. I’m talking about somewhere that it wouldn't be strange for the CEO to say, “Hey, I’m messing around with this, here's what I'm learning, and I really want you guys to all get your hands on it. And we are not here to remove a lot of jobs.”

Now, you have to be honest about that and believe that AI is not meant to go around employees. Rather, it’s meant to go through them, and if you can augment what they do, that will drive market cap, revenue, or whatever is the most important thing for your company.

What are the top three pieces of advice you have for C-suite executives that want to get started?

The number one thing is to get training on it.

Number two is understanding that the top level needs to understand what this is. It's not enough to delegate it. If the C-suite doesn't understand what change at scale looks like, then they will never be able to create the benchmarks that they need to truly change their organization. I call it “walking the factory floor” to learn what it looks like for people in HR or legal or sales to actually do that job and what it looks like sped up.

The third thing is that you can say all you want, “Hey, people, just get on there and practice.” But people don't know what to practice. So, I think that you have to talk to folks and get them inspired to do it. Show them that this is not a work thing — this is a life thing. Nobody takes home a big CRM and uses it to make their life better at night. But this thing actually makes every part of your life better.

About Conor Grennan

Conor Grennan is Chief AI Architect at NYU Stern School of Business, and a New York Times and #1 International bestselling author. He is also CEO and Founder of AI Mindset, an AI consulting and training company.

Training business brains to use GenAI