A top Citi banker warns AI could mean a 'tragic end' for capitalism if we don't act now

For longtime Citi banker Jay Collins, the rise of AI and robotics represents a threat to the future of capitalism that can't be ignored.

Collins, who is Citi's group chairman of the public sector, has spent more than three decades advising government officials during times of crisis and financial duress.

With AI and robotics, Collins sees a challenge to capitalism that policymakers and business leaders must address head-on.

"We have to tweak it, remodel it, remake it to allow for this, just like we did during the Industrial Revolution," Collins told Business Insider. "Unless you're going to go to an authoritarian-type capitalist regime, we've got to figure out how to make this work."

Collins spoke to Business Insider as part of our wider series on the Future of Capitalism. He outlined the impact robotics could have on the workforce, why the middle class is at risk, and how a "productivity dividend" could be the answer.

The following is a condensed, lightly edited version of the conversation.

Dan DeFrancesco: How are you seeing AI impact capitalism, both now and in the future?

Jay Collins: The first wave is cognitive jobs. It starts with white-collar, not blue-collar. It's impacting the college kids who thought a college degree was everything. The graduate students who said, 'If I can just learn to code, I'm safe.' It's hitting software, financials, media, consultants, accounting, and lawyers.

That wave of disruption is also interesting because it should be framed in the context of what's happening in the K economy.

You have half our population at the bottom. You have this 10% at the top, with a net worth of $2 million or more.

The middle of the K has stagnated. They haven't really been hit yet. But they don't have equity assets. So what happens over time as you move into the second wave? You have the 10% who are not only high-wage earners, but more importantly, the largest participants in the equity market. So they're basically riding the benefits of the wealth creation and increasingly separating themselves from those who are not participating, including the middle class.

What does that next wave bring?

Collins: If you look at it as four phases of AI. First, there's the anticipation of gen AI, knowing it was coming forever, and then it hits. That's really the first tier of AI.

The next stage that we're in now is agentic AI. Think of it as going from analysis and research to doing. An agent can act.

To then physical AI. And physical AI is, think of it as a Jensen Huang term: What happens on the manufacturing floor when you have AI-plus-robotics at scale.

And then of course, you have the last phase, where the future of capitalism is most challenged: AGI, the super-intelligence phase.

We still probably have 500,000 or more unfilled manufacturing jobs, so you're not seeing that blue-collar impact yet.

Depending on who you listen to, you're probably talking 2028 or 2029 for that next wave of job disruptions.

We're not there yet, where we will rush to put AI and robotics into manufacturing facilities. That's when you really get to the blue-collar job disruption.

How do you see some of the current policy makers, like the Federal Reserve, reacting to AI?

Collins: At a certain point, the Fed is pushing on a string. Its ability to influence this is increasingly limited. If you simply respond by trying to push money into the system, you can just create inflation. So you start with one premise: big-picture economic numbers now are not only less meaningful; in this age, they're dangerous. In fact, I would say those data points for the overall economy are misleading, and AI and robotics really mean the death of those.

You'll start to see the value of labor drop. Therefore, what is available to a central banker in this world to respond to that? It throws the policy tools back into the corner of fiscal tax revenue, wealth distribution, social benefits, and social safety nets.

You also see China as an important part of the equation.

Collins: The smartest techies I've talked to — the leaders of tech, AI, and robotics — most of them would slow this down if they could, because we're not ready. They're screaming that we're not ready. And they see that we haven't really tested how to come to grips with it.

Why are they saying they can't? They're saying they can't because of China. They're saying they can't because of the geostrategic component.

There is a geostrategic competition happening in the world that makes it a race no one can afford to slow down.

What policies could be put in place to get ahead of this?

Collins: I think it's fair to say that everybody hates universal basic income. On the left and on the right. It gets kind of named and tagged. It's easy to hate. There are phases of it. There are degrees of it. But it is often explained by some of the smartest thinkers since the earliest days of the concept of universal basic income as the most necessary response.

So, I would say first and foremost, you've got to come together and figure this out because no response is actually a tragic end to capitalism and potentially to democracy. I don't want to predict how fast that is, but at a certain point, this disruption requires a response.

The response mechanisms sound to the right like socialism. There's no job. I'm just basically giving a universal income. The right would say, first of all, that's a massive fiscal problem. We can't afford that fiscally. But two, what happens to the incentive structure that has made capitalism great?

If you put on a UBI that disincentivizes everything, then you fundamentally destroy what has driven us to how we got here in the first place.

So what's the fix?

Collins: Let's call it a productivity dividend, because UBI has so much baggage. And if you're talking about how to distribute that productivity dividend, I can start putting it on at very low levels to avoid materially impacting someone's self-esteem with work and their incentive to work.

And there's a conservative thought process that says, as you do that, take out all of the flawed benefits that are happening in the meantime, and therefore, this should be a replacement.

You can phase the level in and see what that does to the incentive structure. Do people actually feel safer taking risks? Some would argue that. Others would say the evidence shows that, once you cross a certain threshold, you've been disincentivized. But we've got to experiment now aggressively with this because we've got to see what can be done.

Start by cleaning up a lot of the benefit programs so that this becomes, at the bottom, a layer upon which to build over time as we watch this unfold. Then you can supplement it for core categories as necessary. But there is no consensus in the policy community as to how to do this.

The real challenge will be when you get into what that looks like: What is the benefit payment program, and how does it impact the economy?

There's also the question of where you get the money? How are you raising it? There are all kinds of ideas from taxing robots, taxing AI, and taxing assets. Depending on how far left, center, or right you are, you have different views on what that'll do. From taxing billionaires to adjusting corporate tax rates to targeting the industries driving the AI and robotics evolution down to their facilities and/or the robots themselves.

Why could this be a good fix?

Collins: Think about what it would mean for us to have some mandatory equity. Wealth creation either through a sovereign wealth fund redistribution mechanism or a complete redesign of the welfare system. So it's not just cash. There's actually a participation in wealth.

I do think we're going to need a congressional commission. I was a senior advisor on the Baker-Hamilton Commission for Iraq. There have been other successful commissions in our country's history. Democrats and Republicans, former cabinet leaders, labor leaders, and CEOs of businesses and Big Tech, who are able to study deeply and come up with recommendations so that the country at least has a foundation for where the compromise lies and what could be done.

Even if it took a year or two to get there, you may not make enough of a difference early on in this first wave, but you would get there.