Collective Prosperity

Key Points:

• The story of prosperity in an AI age depends on whether we use AI to complement or compete with human skills, and how we define knowledge discovered by machines.
• Cultural factors determine if AI leads to shared prosperity through productivity gains or zero-sum competition over declining careers. Economic growth requires productivity growth.
• The dominant narrative that AI inevitably lifts productivity overlooks complex realities. A bias for automation for efficiency underestimates context and neglects the human contribution that resist capture by automation.
• If adopted as a collaborative tool rather than replacement, AI can create new areas of human expertise to mitigate skill saturation. The goal should be to focus on having AI tackle previously unsolvable and emergent problems.
• The solution requires seeing AI as amplification rather than automation, combining strengths of human reasoning and AI's high-dimensional knowledge.

The story of prosperity in an AI age remains unwritten. Yet the way we talk about AI often fixates narrowly on AI’s impact on work, asking if machines will displace human work. We believe the more pressing questions are these:

• Will we use AI to complement or compete with our core skills?
• How do we define knowledge when machines know things we can't?
• How does our culture evolve as intelligent, autonomous machines consume more of what we consider to be uniquely human capabilities?

We often think of AI as standalone agents with intelligence of their own. But it's not the only way to think about it. Psychology professor Alison Gopnik proposes reframing AI as "cultural technologies”. She thinks of generative AI as tools which harness our collective knowledge and capabilities that have accumulated over generations.

We agree. How AI helps or hinders us will depend on the culture in which it is adopted. If we collectively get this right, people may flourish with AI. What cultural factors need to align for shared prosperity rather than an outcome of zero-sum competition over dwindling career prospects in an automating world?

As flawed as economic measures are when applied to something as all-encompassing as “human flourishing”, economic growth matters. The future of economic growth is largely going to be about productivity growth. Curiously, even with the staggering innovations in technology, most advanced economies have a problem: stagnant productivity growth. According to Erik Brynjolfsson and Gabriel Unger: “More than any other factor, productivity—output per unit of input—determines the wealth of nations and the living standards of their people.”

Our biggest problems—environmental catastrophe, healthcare, poverty—are far more manageable with higher productivity (and good jobs). But here's a troubling truth: how we adopt AI is a function of narrative and the dominant narrative is in trouble. In the USA, the default narrative is that AI inevitably leads to higher productivity and economic growth, benefiting everyone in the long run. Acemoglu points out that the reality is more complex. A narrow focus on automation often overestimates of AI’s capabilities.

“The path that leads to a worse future is the one of least resistance and results in low productivity growth, higher income inequality, and higher industrial concentration.”—Erik Brynjolfsson and Gabriel Unger

This matters because jobs comprise diverse, contextual skill sets, not singular tasks. When businesses automate with an efficiency mindset, they overlook indispensible human contributions. Work is reshaped around machine limitations, creating “bullshit jobs” where people fill residual gaps that only humans can occupy. Productivity languishes. Acemoglu and Johnson attribute lackluster economic growth partially to this “so-so technology” dilemma.

David Autor warns of the unique challenge AI poses to human roles: as AI meets the demand for certain skills, the value and compensation for these skills may collapse, leaving many previously-valuable human skills commercially obsolete. If adopted as a collaborative tool, rather than a replacement, AI will help us create new human areas of expertise and productivity and mitigate some of the effects of skill saturation.

The ultimate goal of enhancing productivity, both individually and system-wide, is to tackle more challenging problems—those that are hard, complex, previously unsolvable, or emergent. A genuine AI utopia begins with tangible, positive scientific advancements, like new materials and discoveries made possible by the deep combinatorial search capabilities of modern machine learning.

AI can open and explore spaces of possibility. In high dimensional spaces, AI will always outperform humans. Navigating new spaces of possibility will require both human mechanistic coarse grained reasoning and machines’ high dimensional statistical knowledge. AI-enabled drug discovery is a case in point: AI's ability to sift through high-dimensional data allows it to identify novel drug candidates that humans might overlook due to the complexity and sheer volume of information. This combines human causal insight on disease mechanisms with AI's computational power to predict molecular behavior.

But greater prosperity is more than advances in data-driven discovery: it will require radical ideas, which are hard to introduce. Whether in science or economics or business, we may enter a world where recursive knowledge from AI is possible and is in fact better than what people can do. It may take quite some time for ideas generated by AI to be fully accepted in our broader socio-economic system.

The solution requires seeing beyond automation to amplification. If society chooses to complement strengths between humans and machines, more dynamic partnerships become possible.