Keeping an Eye on AI

Content

One of the foremost thinkers in applied artificial intelligence, Zack Kass, will present his keynote address Monday morning during ATD26. He is author of The Next Renaissance: AI and the Expansion of Human Potential.

One of the foremost thinkers in applied artificial intelligence, Zack Kass, will present his keynote address Monday morning during ATD26. He is author of The Next Renaissance: AI and the Expansion of Human Potential.

Content

As an AI strategist, executive business advisor, researcher, and the former head of Go To Market at OpenAI, Kass has a mission to ensure individuals, businesses, and governments are active participants in the AI-powered future by making the technology both understandable and actionable.

As an AI strategist, executive business advisor, researcher, and the former head of Go To Market at OpenAI, Kass has a mission to ensure individuals, businesses, and governments are active participants in the AI-powered future by making the technology both understandable and actionable.

Content

He answered questions from Conference Daily prior to his keynote address.

He answered questions from Conference Daily prior to his keynote address.

Content

Can you tell us about your work with the University of Virginia regarding AI and what your research has found to date?

Can you tell us about your work with the University of Virginia regarding AI and what your research has found to date?

Content

The work I’m doing with the University of Virginia looks at the social and economic consequences of AI. I tend to organize the conversation into three categories: neutral forces that will shape the transition, risks we should take seriously, and reasons for optimism.

The work I’m doing with the University of Virginia looks at the social and economic consequences of AI. I tend to organize the conversation into three categories: neutral forces that will shape the transition, risks we should take seriously, and reasons for optimism.

Content

Let me start with the neutral dynamics, because those are the ones we should probably expect regardless of where the technology ultimately lands.

Let me start with the neutral dynamics, because those are the ones we should probably expect regardless of where the technology ultimately lands.

Content

One is political protection. Every major technological transition produces pressure on governments to slow disruption. Labor groups organize, industries lobby, regulations emerge, and adoption spreads more slowly than the technology technically allows. Markets tend to move quickly; institutions move slowly. That friction is not unusual—it’s part of how societies absorb technological change.

One is political protection. Every major technological transition produces pressure on governments to slow disruption. Labor groups organize, industries lobby, regulations emerge, and adoption spreads more slowly than the technology technically allows. Markets tend to move quickly; institutions move slowly. That friction is not unusual—it’s part of how societies absorb technological change.

Content

Another neutral dynamic is what I sometimes call the apprenticeship problem . Historically, many professions trained people by having them do smaller pieces of the job before graduating to the larger responsibility. Analysts built the spreadsheets before they ran the deals; junior lawyers reviewed documents before arguing the case.

Another neutral dynamic is what I sometimes call the apprenticeship problem. Historically, many professions trained people by having them do smaller pieces of the job before graduating to the larger responsibility. Analysts built the spreadsheets before they ran the deals; junior lawyers reviewed documents before arguing the case.

Content

If AI begins performing many of those early tasks, fewer people will learn through that traditional path. That sounds alarming, but it may also compress career timelines. More people may move quickly into higher-level work instead of spending decades doing their time.

If AI begins performing many of those early tasks, fewer people will learn through that traditional path. That sounds alarming, but it may also compress career timelines. More people may move quickly into higher-level work instead of spending decades doing their time.

Content

And of course there is the psychological side.

And of course there is the psychological side.

Content

Most people are perfectly happy for technology to automate someone else’s job—just not their own.

Most people are perfectly happy for technology to automate someone else’s job—just not their own.

Content

That instinct is deeply human, but it is also rational. At a societal level, most people understand that automation tends to increase prosperity. When productivity rises, economies grow, prices fall, and living standards improve. The benefits are broad and diffuse.

That instinct is deeply human, but it is also rational. At a societal level, most people understand that automation tends to increase prosperity. When productivity rises, economies grow, prices fall, and living standards improve. The benefits are broad and diffuse.

Content

The disruption, however, is concentrated. If your specific role is the one being automated, the collective economic gain can feel abstract compared with the immediate loss of status or income. That tension—between diffuse societal benefit and concentrated individual disruption—shows up in almost every automation debate.

The disruption, however, is concentrated. If your specific role is the one being automated, the collective economic gain can feel abstract compared with the immediate loss of status or income. That tension—between diffuse societal benefit and concentrated individual disruption—shows up in almost every automation debate.

Content

That leads to one of the real risks, which I’ve written about extensively: the distribution problem.

That leads to one of the real risks, which I’ve written about extensively: the distribution problem.

Content

AI may expand the economic pie dramatically. The question is how the gains are distributed. If productivity growth concentrates too narrowly—among a handful of firms, regions, or workers—the transition could feel destabilizing even if the technology itself is creating enormous wealth.

AI may expand the economic pie dramatically. The question is how the gains are distributed. If productivity growth concentrates too narrowly—among a handful of firms, regions, or workers—the transition could feel destabilizing even if the technology itself is creating enormous wealth.

Content

There are also some early signals in the labor market worth paying attention to. Recent work by Erik Brynjolfsson and others suggests that hiring for certain entry-level knowledge roles—particularly among new graduates—may already be softening in response to AI capabilities. It’s too early to draw sweeping conclusions, but it’s a development worth watching.

There are also some early signals in the labor market worth paying attention to. Recent work by Erik Brynjolfsson and others suggests that hiring for certain entry-level knowledge roles—particularly among new graduates—may already be softening in response to AI capabilities. It’s too early to draw sweeping conclusions, but it’s a development worth watching.

Content

Another issue is what I call identity displacement .

Another issue is what I call identity displacement.

Content

For decades, many professionals have defined their value around specific cognitive outputs—the memo, the analysis, the model, the presentation. When machines begin performing those tasks well, people can experience a kind of professional disorientation. Even if the job itself still exists, the part that once anchored their sense of value suddenly feels less scarce.

For decades, many professionals have defined their value around specific cognitive outputs—the memo, the analysis, the model, the presentation. When machines begin performing those tasks well, people can experience a kind of professional disorientation. Even if the job itself still exists, the part that once anchored their sense of value suddenly feels less scarce.

Content

That psychological adjustment may turn out to be one of the most important aspects of the transition.

That psychological adjustment may turn out to be one of the most important aspects of the transition.

Content

But there are also reasons for optimism.

But there are also reasons for optimism.

Content

Historically, economies have repeatedly discovered entirely new forms of work as technology changes the production frontier. Entire industries that employ millions of people today—software development, digital advertising, cloud infrastructure—barely existed a few decades ago. The discovery of new work has been a persistent feature of technological progress.

Historically, economies have repeatedly discovered entirely new forms of work as technology changes the production frontier. Entire industries that employ millions of people today—software development, digital advertising, cloud infrastructure—barely existed a few decades ago. The discovery of new work has been a persistent feature of technological progress.

Content

And there is a deeper possibility as well. If AI meaningfully expands productivity, it moves us closer to solving what John Maynard Keynes once called the economic problem —the challenge of producing enough goods and services to meet human needs.

And there is a deeper possibility as well. If AI meaningfully expands productivity, it moves us closer to solving what John Maynard Keynes once called the economic problem—the challenge of producing enough goods and services to meet human needs.

Content

In 1930, Keynes wrote that humanity might eventually reach a point where the economic problem is largely solved and people must confront something more profound: how to use their time meaningfully once survival is no longer the central constraint.

In 1930, Keynes wrote that humanity might eventually reach a point where the economic problem is largely solved and people must confront something more profound: how to use their time meaningfully once survival is no longer the central constraint.

Content

Frankly, I think our understanding of “work” and a “job” will change so fundamentally that we should expect a spiritual awakening on the other side of it.

Frankly, I think our understanding of “work” and a “job” will change so fundamentally that we should expect a spiritual awakening on the other side of it.

Content

What skills are required of humans in the age of the AI Renaissance?

What skills are required of humans in the age of the AI Renaissance?

Content

One way I think about the transition is through my theory of unmetered intelligence.

One way I think about the transition is through my theory of unmetered intelligence.

Content

For most of human history, intelligence was scarce. If you wanted analysis, research, writing, or strategic thinking, you had to hire highly trained people and pay for their time. Cognitive labor was expensive and unevenly distributed.

For most of human history, intelligence was scarce. If you wanted analysis, research, writing, or strategic thinking, you had to hire highly trained people and pay for their time. Cognitive labor was expensive and unevenly distributed.

Content

AI begins to change that.

AI begins to change that.

Content

Certain forms of intelligence—summarizing information, drafting documents, analyzing data, retrieving knowledge—are becoming extremely inexpensive and widely available. In that sense, intelligence begins to behave less like a scarce commodity and more like a utility (something closer to electricity or bandwidth).

Certain forms of intelligence—summarizing information, drafting documents, analyzing data, retrieving knowledge—are becoming extremely inexpensive and widely available. In that sense, intelligence begins to behave less like a scarce commodity and more like a utility (something closer to electricity or bandwidth).

Content

When that happens, the structure of advantage shifts.

When that happens, the structure of advantage shifts.

Content

For a long time, professional success often depended on having the right answers. In a world of unmetered intelligence, answers become easier to generate. What becomes more valuable is asking the right questions—framing problems clearly, understanding what actually needs to be solved, and knowing which answers are worth trusting.

For a long time, professional success often depended on having the right answers. In a world of unmetered intelligence, answers become easier to generate. What becomes more valuable is asking the right questions—framing problems clearly, understanding what actually needs to be solved, and knowing which answers are worth trusting.

Content

Another durable skill is learning how to learn.

Another durable skill is learning how to learn.

Content

The tools will continue to evolve quickly. Techniques that feel advanced today will look primitive in a few years. People who succeed in that environment are rarely the ones who mastered a single method; they are the ones who can repeatedly absorb new tools, experiment with them, discard what doesn’t work, and move on.

The tools will continue to evolve quickly. Techniques that feel advanced today will look primitive in a few years. People who succeed in that environment are rarely the ones who mastered a single method; they are the ones who can repeatedly absorb new tools, experiment with them, discard what doesn’t work, and move on.

Content

And there is a more human quality that matters as well: pain tolerance.

And there is a more human quality that matters as well: pain tolerance.

Content

Periods of rapid technological change can feel disorienting. Workflows break. Expertise decays. The thing you spent 10 years getting good at suddenly becomes easier for a machine to replicate.

Periods of rapid technological change can feel disorienting. Workflows break. Expertise decays. The thing you spent 10 years getting good at suddenly becomes easier for a machine to replicate.

Content

People who navigate that transition well tend to tolerate that discomfort. They experiment, they rebuild their workflows, and they accept that professional identity may need to evolve more than once.

People who navigate that transition well tend to tolerate that discomfort. They experiment, they rebuild their workflows, and they accept that professional identity may need to evolve more than once.

Content

What do you hope ATD26 attendees will take away from your speech?

What do you hope ATD26 attendees will take away from your speech?

Content

Mostly perspective.

Mostly perspective.

Content

Right now, many conversations about the future—not just AI conversations—are saturated with a kind of ambient dread. Climate anxiety, geopolitical instability, economic uncertainty; there is a general sense that the world is becoming more complicated and less predictable.

Right now, many conversations about the future—not just AI conversations—are saturated with a kind of ambient dread. Climate anxiety, geopolitical instability, economic uncertainty; there is a general sense that the world is becoming more complicated and less predictable.

Content

AI has simply become the newest vessel for that feeling.

AI has simply become the newest vessel for that feeling.

Content

My goal is not to dismiss those concerns. Technological transitions are disruptive; they always have been.

My goal is not to dismiss those concerns. Technological transitions are disruptive; they always have been.

Content

What I try to offer instead is historical context and a framework for thinking about these changes in a way that is grounded in evidence rather than panic.

What I try to offer instead is historical context and a framework for thinking about these changes in a way that is grounded in evidence rather than panic.

Content

The conversation about job loss, for example, is probably permanent. As long as technology continues advancing, societies will continue debating whether machines are replacing too much human labor.

The conversation about job loss, for example, is probably permanent. As long as technology continues advancing, societies will continue debating whether machines are replacing too much human labor.

Content

But scary does not necessarily mean bad.

But scary does not necessarily mean bad.

Content

Many of the technologies that ultimately improved human living standards the most were deeply unsettling when they first appeared.

Many of the technologies that ultimately improved human living standards the most were deeply unsettling when they first appeared.

Content

What matters is how societies respond.

What matters is how societies respond.

Content

If there is a theme running through my work, it is the idea of constructive responsibility as an alternative to ambient dread. Technology expands the frontier of what is possible; humans decide what to build within that expanded frontier.

If there is a theme running through my work, it is the idea of constructive responsibility as an alternative to ambient dread. Technology expands the frontier of what is possible; humans decide what to build within that expanded frontier.

Content

And those decisions—economic, political, and cultural—are still ours to make.

And those decisions—economic, political, and cultural—are still ours to make.

Content

Read more about ATD26 at conferencedaily.td.org .

Read more about ATD26 at conferencedaily.td.org.