Editors note: This interview was conducted in August 2023.
It’s palpable that we’re living in an era of irrepressible change. You can trace it back to the turn of this century or earlier, but it’s become that much clearer that we are witnessing a great many transformations just within the last few years—or even the last few months.
When OpenAI’s ChatGPT launched at the end of November 2022, it rightfully caused a sensation; many quickly realized that the technology was carrying us to an unexpected place and at a pace faster than we could have possibly imagined.
But Zack Kass, who has spent 14 years in AI, most recently leading sales, solutions, and partnerships at OpenAI, saw the start of this sweeping change coming a long time ago. Regarding applied AI, he’s admittedly spent more time with the business side of the tech than the majority of his contemporaries. “I know how the world adopts AI better than most people,” says the Santa Barbara, CA native. Accordingly, Kass describes himself as an AI Futurist.
Kass’s unique position, history, and experience with AI will shine when he keynotes at the 2023 McIntire Fall Forum, “The AI Revolution: Exploring Possibilities & Challenges,” taking place on Oct. 20 at 10:30 a.m. at UVA’s Old Cabell Hall Auditorium. We recently spoke with Kass about the many ways AI is set to upend and, yes, better our world. We discussed where we currently stand in what appears to be a lengthy journey, one capable of birthing sudden and continuous changes with the power to positively impact business and society at large.
What’s the most important thing about generative AI that most people should know but don’t?
Kass: We should define “generative AI.” For the record, Sam [Altman, OpenAI CEO] doesn’t use “generative AI.” Almost no one at OpenAI does. “Generative AI” is this term that the industry has sort of co-opted. “Generative AI” is actually kind of a funny term, because it’s a lot more than just generative. GPT is a model, but there are a lot of other models in the canon of what is now modern AI.
That that being said, regarding what the industry thinks of as generative AI, its most important moments have happened: the invention of the transformer, the explosion of data, the zero interest rate environment that led to this incredible amount of accessible compute. These are the seminal moments that have led to where we are today, but where we’re going is arguably even more exciting. We’re still in the curve of a logarithmic vector that is going to explode in pretty incredible ways over the next five or 10 years. Maybe faster, but probably not much faster.
The most exciting thing is how fast the world has updated to the potential of this technology. And the reason I say that is GPT 3.5 existed six months before ChatGPT was released, which means that anyone in the world could have built ChatGPT almost as it exists today without any real, noticeable differences. ChatGPT is a slightly more aligned model to discussion, to discourse. That’s to say that the technology has sort of proven “if you build it, they will come” is not inherently true. There’s a lot of technology that actually takes people updating. And I think the most interesting thing in generative AI is the public market updating to the potential of this technology.
So would you say that the most important thing that people don’t realize about AI is that it’s still very early times?
Kass: The most important thing that people don’t know is that it is early on the J-curve, and that we are scratching the surface in terms of modality. It is expansive in terms of performance, as GPT 5 will be much better than 4, and in terms of breadth, computer vision is coming very soon. Both listening and translating live are only going to get better. The model speaks 140 languages today; it’ll probably speak a thousand in a year. We’re going to resurrect dead dialects and invent new languages. I don’t think people have fully grasped all of that yet.
Where do you see the most potential for growth with AI-related business applications in the coming months in terms of trends?
Kass: Probably the three most important trends are 1) public perception, 2) public policy, and 3) application adoption.
There’s this weird, muddled public perception right now. We’re in the middle of this interesting backlash with Hollywood, Washington, and various nation-states; Italy had that ChatGPT ban. People aren’t sure how to feel about it, and I think that’s going to be coming into much greater focus over the next six months. I think it could break both ways. This isn’t to say that I think people will like it more in six months, but I think there will be a much stronger understanding in six months of how this is going to shake out.
That’s going to be particularly important, leading up to a presidential election, because there is no doubt that AI will be a presidential election topic. It will come up in a debate.
Adoption is pretty interesting, because right now, there’s this leading indicator. All of these companies are building AI-powered applications that are all about to start maturing out of beta and alpha into GA [general availability]. When that happens, people are going to start to notice pretty meaningful improvements in, for example, customer support, sales, and marketing personalization. It’s going to happen slowly and then suddenly. And it’s going to be incredible, because it’s going to lead to a pretty inherent deflationary element to AI. Cost of goods and services is going to go down. Customer experience is going to go up. Then people are going to start to settle into the reality of what the new world might look like, which is a far more automated one. Seamless in many ways, and sort of unnerving in others.
How do you see AI’s role in supporting companies to achieve their goals?
Kass: OpenAI, Anthropic, Google DeepMind, and Microsoft, as the private, non-state actors in the category, are all very aware of the labor and economic implications of this technology and of the associated non-economic risks inherent in bias and safety. There are huge teams at every one of these companies designated to uphold safety and AI responsibility standards, and they are now working in concert, acting as an industry group.
It is also true that there is only so much that can be done to “protect jobs,” I expect. But I take a very rosy outlook on this future and expect there will be explosions of new jobs that we can’t even imagine over the next five to 10 years.
But even still, it’s hard not to think that in the next 10 to 15 years we will see a shortfall of jobs. But it may not be bad that we only have 30 hours of work to do every week. No one at the McIntire School is going to argue that people don’t work enough. People are working way too much. Then the question is “What’s the sweet spot? Should people all get paid the same salary if we work 30 hours per week? What about 20?”
There’s a point at which we’ll start to get paid less because we’re doing less work. But maybe that’s okay. For example, I love playing volleyball, and I would take a pay cut to play volleyball two hours a day. When I talk to groups, I talk about looking forward to a world where we recreate more. I ask everyone, “Show of hands; who would rather be remembered as a great friend or family member versus a great professional?” Not a single person wants to be remembered only as a great professional. No one. Now I’m trying to practice what I preach by finding balance in my life and starting to rediscover an identity as someone who is a family member and a friend and volleyball player. I grew up playing volleyball and am just rediscovering the things in my life that I love.
So what if we imagine a world where that’s going to happen? That doesn’t actually change the fact that there will be short-term pain with respect to the labor market. I spend time talking to companies about how to manage that.
Sam is also pretty outspoken about this. He says that nothing is going to get replaced anytime soon. These are all additive. If anything, we’re going to have to create more jobs because companies are going to be able to buy and sell more wares because cost of services is going to come down.
Do you see a way for companies to avoid creating an impersonal environment for business if many of these jobs and functions are replaced by AI? How can they incorporate the technology without dehumanizing or automating relationship-based exchanges that are often at the core of business?
Kass: Morgan Stanley has taken a really beautiful tack, which I think most financial services are going to emulate. And certainly, most professional services, too. They have built a machine that provides a sort of value-added service to their wealth managers, with the goal not being effectiveness but efficiency.
If a Morgan Stanley wealth adviser can use AI to manage, for example, $50 million dollars instead of $25 million dollars, then maybe a number of things happen: First, Morgan Stanley grows its assets under management. That’s great for their business. But second, presumably the wealth adviser is now doing more high-level strategic work and less grunt work, which improves their job satisfaction, etc., etc. It also is proven now to reduce number of errors. So effectiveness is becoming a core piece of this technology.
My dad’s a Morgan Stanley wealth client. If he emails his guy at 9 p.m., his guy might actually feel compelled to send an email. And at 9 p.m., maybe he’s had a couple of glasses of wine. Compliance is a is a hairy thing in wealth advisory—it’s really hard to be 100% compliant. Now, this machine says, “Hey, this thing that you’ve drafted actually has these errors.” Maybe it auto-drafts the response to my dad, and his wealth adviser reviews it instead of drafting it himself. That’s not going to cut Morgan Stanley jobs; that’s going to increase the amount that Morgan Stanley can manage.
What’s really exciting is that maybe that actually contributes to deflationary cost of service at Morgan Stanley, so that they can start charging a lower rate to their clients, and that makes wealth management more accessible for younger clients who don’t want to pay, or who are used to paying only a robo-adviser fee. To me, that is where this is ultimately going, which is that things that are luxuries today will be staples in the future. Great education, great financial services, access to medical care, all these things that are reserved for the wealthy today will start to become more and more accessible.
But again, there’s no doubt that companies need to be really mindful if they’re building AI to eliminate jobs or to improve the efficacy and efficiency of their employees’ work. And for 99% of the companies that we work with, the latter is true, and they are very outspoken about that.
Let’s talk about one of the coolest things you’ve seen so far from an application of AI.
Kass: The application that I love was when Aaron Levie launched Box Intelligence. Cloud storage is so boring. There’s nothing interesting about cloud storage. Aaron always had this vision that you would put everything in Box, and it would make sense of all the stuff. They tried so many different ways to aggregate the information because, in theory, if you put your life in Box and it can organize and index it, and you can search it infinitely, and you can ask it the interesting questions, then suddenly cloud storage is super interesting. Now it’s aware. Now it’s like self-storage except that it’s self-organizing and self-reporting.
You can just say, “Show me all contracts that are expiring this year.” Or “Show me all contracts signed by this person who’s no longer at the company who we worry wasn’t reviewing contracts effectively.” Or “Show me all contracts that have conflicting clauses.” This is very powerful, very real stuff that’s about to hit the market.
And it dawned on me that we’re about to go through this age when all the software that we’ve basically been building since 2000 will finally deliver on all the worker productivity gains that we were promised. We’ve effectively been worker-productivity flat since 1992 in this country. Suddenly, we’re about to see incredible moments come to fruition with all of the software companies that have been promising magic in our lives for a long time.
It’s going to happen fast. Software is about to become magical. It will be this incredible moment when we realize that software is really value additive, and everyone’s about to get access to that.
You use the title “Futurist” [Kass’s former title at OpenAI], which has spurred a lot of conversation around the School. Can you explain what that means within a business context?
Kass: Futurist is an elevated title that I’m sometimes sheepish to use, because I don’t have a Ph.D. and I’m not a scientist, but I’ve genuinely spent as much time as anyone in this world helping people adopt modern AI in safe, practical ways.
Having built OpenAI go-to-market, sales, marketing, partnerships, solutions, and programs, the thing that I get most excited about is helping leaders imagine what the future will look like. I’m not [American computer scientist, author, inventor, and futurist] Ray Kurzweil, but I have pretty good opinions on how the world will adopt and apply this technology. I’ve seen a lot, and I think I’m as equipped as anyone in the world to sort of predict how this will all map together in practical ways.
So now I spend most of my time talking to heads of state, policymakers, regulators, business executives—talking to anyone who will listen, frankly about the safe and effective application of this technology now and in the future. I have political aspirations because I just think the world is about to change so much. And the most dangerous thing we could do is build a self-fulfilling pessimistic prophecy, but I really don’t want to see that happen.
Right now is the time for future generations to be hopeful. The thing that gives me the most concern and pause is just inherent pessimism, especially with Gen Z. There’s a lot of reason for people to look at the world sideways, but we have this unique opportunity right now to be optimistic about this future.
Because there are billions of people on Earth right now who have never seen a doctor. There are billions of people on Earth who don’t have access to what we’d consider effective quality education. That’s the problem here. The problem is not whether Hollywood is getting automated, or if Tom Cruise’s likeness is going to be imitated in 20 years; the problem is that there is an incredible amount of suffering in the world that we need to solve. We have the opportunity to do that right now.
It’s such an amazing time to be alive. In our lifetime, we’re going to see AGI [artificial general intelligence] and interstellar travel. That’s wild.