The AI industry is split between two radically different visions of the future: tech leaders predicting near-science-fiction growth, and economists expecting modest productivity gains that barely move the needle on overall economic progress. This isn't just academic disagreement; it's shaping how companies invest billions, how governments set policy, and whether we should expect transformative change or incremental improvement from artificial intelligence over the coming years.

What's Driving This Massive Gap in AI Growth Predictions?

On one side sits what observers call the "San Francisco Consensus," where tech leaders and venture capitalists see AI as a revolutionary force that could accelerate productivity, solve complex problems, and unlock entirely new industries. On the other side, institutions like the Federal Reserve and Congressional Budget Office continue to forecast sub-two percent economic growth, while skeptics like Nobel laureate Daron Acemoglu argue that AI will deliver only minimal productivity gains in the real world.

This divide isn't new. Throughout history, transformative technologies have sparked similar debates. The internet was dismissed as a fad by some economists, while others predicted it would solve every problem. The reality fell somewhere in between. But with AI, the stakes feel higher because the potential applications span nearly every sector of the economy, from healthcare to manufacturing to knowledge work itself.

How Are Researchers Actually Measuring This Disagreement?

Survey data from the Forecasting Research Institute reveals just how wide the gap has become between optimistic and cautious views on AI's economic impact. Rather than a gradual spectrum of opinion, the data shows a clear bifurcation: believers in transformative AI growth on one side, skeptics expecting modest improvements on the other. This polarization matters because it influences everything from startup funding to corporate strategy to regulatory decisions.

The disagreement hinges on a practical question: will AI breakthroughs in the lab translate into meaningful productivity gains in real-world workplaces and industries? Tech leaders point to rapid improvements in AI capabilities, while economists note that previous technological revolutions took decades to show up in productivity statistics. The question of whether AI will be different remains genuinely unresolved.

Ways to Think About the AI Growth Question More Clearly

• Distinguish Between Capability and Adoption: An AI system can be incredibly capable in the lab but still struggle to integrate into existing workflows, regulatory frameworks, and organizational structures. Real productivity gains require both breakthrough technology and successful deployment at scale.
• Look at Historical Precedent With Skepticism: Previous technologies like electricity and the internet did eventually transform economies, but the process took 20 to 40 years. Assuming AI will be faster requires evidence, not just optimism.
• Examine Sector-Specific Evidence: Rather than debating AI's impact on "the economy" broadly, focus on specific industries where AI is already being deployed. Healthcare, finance, and manufacturing offer real-world test cases for whether productivity gains materialize.
• Consider the Role of Complementary Investments: Technology alone doesn't drive growth. Education, infrastructure, and institutional adaptation matter just as much. AI's impact depends partly on whether society invests in these complementary areas.

What Does This Debate Tell Us About Progress More Broadly?

The AI growth disagreement reflects a deeper tension in how we think about technological progress. One perspective, articulated by researchers at McKinsey in their work "A Century of Plenty: A Story of Progress for Generations to Come," argues that we should expect continued dramatic improvements in human welfare over the coming century, just as we've seen over the past hundred years. Life expectancy has nearly doubled, extreme poverty has dropped from around 60 percent to 10 percent, and child mortality has plummeted.

These authors suggest that the only thing holding us back from another century of similar progress may be our own disbelief in the possibility. They introduce the concept of an "empowerment line" alongside the traditional "poverty line," asking not just whether people can avoid extreme deprivation, but whether they have enough security and breathing room to shape their own lives, invest in themselves, and build resilience against shocks.

If this vision is correct, then AI's role becomes clearer: not as a standalone miracle solution, but as one tool among many for expanding human capability and opportunity. The question then becomes whether we'll use AI to focus on essentials like affordable housing, healthcare, transportation, education, and energy, or whether we'll allow it to concentrate wealth and power among a narrow group of companies and individuals.

Why the Geopolitical Context Matters for AI's Future Impact

There's another layer to this debate that often gets overlooked: geopolitical risk. Today's US-China tensions threaten to fragment the global research and talent networks that have driven innovation for decades. Restrictions on talent, research collaboration, and knowledge sharing could quietly undermine long-term technological progress, regardless of how capable individual AI systems become.

History offers a cautionary tale. World War I devastated Britain's educated class, creating a "lost generation" that reduced patent output for decades, especially in breakthrough fields. Global scientific collaboration collapsed, with bridges of international knowledge suddenly burned and productivity falling sharply. Even without outright conflict, geopolitical ruptures can weaken the innovation ecosystem that makes progress possible.

The implication is sobering: AI's transformative potential depends not just on technical breakthroughs, but on whether the world maintains the collaborative networks, talent mobility, and open knowledge sharing that have historically driven innovation. If geopolitical tensions fragment these networks, even the most capable AI systems may fail to deliver the broad-based productivity gains that optimists predict.