AI Unlocks Cognitive Elasticity

Job growth is going to accelerate

The Two-Org Future

Arnold Kling

wrote an essay this week called When AI Collides with Corporate IT. The essay focuses on how corporate IT departments will prevent and hinder the deployment of AI because they will see it as a threat. Specifically, Arnold writes, “Corporate IT (and government IT) will advance one funeral at a time. We will have to wait for AI natives to make it to the executive suite.” I think there are relevant historical analogies, but I also might take a slightly different take on this.

The last time corporate IT faced an existential fork in the road was the shift from on-premise data centers to the public cloud. In some companies, the IT department became a blocker rather than an enabler. This slowed progress. I saw it firsthand when I worked at Microsoft. I met with countless large enterprises to discuss their approach to digital transformation. Many of the IT departments seemed afraid of losing their jobs or worse – totally clueless. If we move to the cloud, what do they need me for?

In many companies, I would meet with the CEO and ask them, “what is your plan to grow the company using digital transformation in the next five years?” Often, the answer was “I’m not sure what the cloud can do. Maybe talk to the IT team.” Then I would go and meet with the IT team and ask the same question, and the answer was usually “we aren’t even sure what the company does. We don’t have a seat at the table for the major strategy decisions like that. It’s our job to order the internet.” In these companies, they were on a path towards doom. No one was responsible for tech-enabled growth, and so no tech-enabled growth happened.

Other companies just got it - and often this came from the board or CEO and became the dominant company culture. These CEOs saw digital transformation as the most important thing on their 5-year plan and they immersed themselves in it. For them, elastic technology capabilities were a strategic weapon. Those companies outperformed their peers.

Netflix moved its entire streaming platform onto AWS, gaining the ability to spin up thousands of servers “within minutes” and push code to production hundreds of times per day. Spotify ditched its self-run data centers for Google Cloud so engineers could “focus better on the core business” instead of babysitting racks. Many other companies embraced the cloud and thrived.

Firms that let IT act as a blocker by hoarding turf, warning of “security risks,” and quietly fearing redundancy all missed that S-curve and have spent the last decade playing catch-up (if they survived at all).

Elasticity of Imagination

I think there is an analogy in AI. At a simple level, there will be companies that are afraid of AI or don’t make a real effort to understand it. They will become the Blockbusters of the new generation. These firms will still adopt AI, but through bolt-on “centers of excellence,” consulting pilots, and dotted-line reporting structures that never reach escape velocity. We can call this the Innovation Prevention Department.

Other firms will embrace AI and will grow faster. This will show up in stock price, earnings multiple, and idea velocity.

But I think there is a deeper level of analysis that we can undertake too. Just as cloud freed companies from the physical capacity constraints of their servers, generative AI will also free us from cognitive capacity constraints. I call it elasticity of imagination.

One human + one large language model can now draft, debug, and deploy software, marketing copy, legal briefs, and customer-success playbooks at a pace that once required a squad of specialists.

The numbers already look like early cloud metrics. In a controlled experiment, developers using GitHub Copilot finished a non-trivial coding task 55.8 percent faster than the control group. Microsoft reports 1.3 million paying Copilot accounts, with the tool generating “nearly half” of their shipped code. That is a 2× productivity shock to white-collar output. This is the human equivalent of doubling the throughput of every factory robot overnight.

What Rational CEOs Do with a 2× Worker

When a piece of capital equipment suddenly doubles its return on invested capital, the CFO doesn’t lay off machines; he buys more of them. By the same logic, enlightened CEOs will hire aggressively into functions where AI superpowers their people. The conventional wisdom that all the white-collar jobs will go away is just totally wrong.

In economics terms, the labor-demand curve shifts to the right: each worker’s marginal product rises, so firms demand more labor at any given wage. History rhymes here. After ATMs proliferated in the 1970s, pundits predicted mass teller layoffs; instead, branch employment rose as lower operating costs allowed banks to open more branches.

Conclusion – The ROI Test

If AI turns every high-potential employee into a compound-interest asset, the winning move isn’t austerity, it’s all-in reinvestment. Companies that treat generative models like the cloud circa 2018 will widen the gap between builders and blockers, not just on the balance sheet but in the war for talent. The scoreboard will be brutal, but that’s the point: capitalism needs visible failures to clear a path for the next great CEO. When productivity doubles, buy more of it, not less. The firms that remember that simple lesson will staff up, accelerate, and leave the blockers behind.