What should humans do next?
It’s perhaps time to reject the idea that humans are social and machines are asocial. As machine capabilities increase and human behaviors evolve, this mode of thinking is a vestige of the 2010s. To remain relevant in a society that will be humans+machines, we need a better understanding of what strengths each of us brings and how to best put them together.
EXPLOSIVE GROWTH: Large-scale AI systems are in an era of explosive growth, demonstrating emergent capabilities impacting the economy. This growth will continue, with some forecasters estimating we may have many features of general intelligence within ten years.
COMPLEXITY AND UNCERTAINTY: AI embodies "creative destruction" (Schumpeter, 1942) - it creates innovation, replaces traditional ways of working, and introduces new uncertainties. Exploring the economic impacts of advanced AI systems is a complex process, given the rate of change and our uncertainty that advanced systems will be closely aligned with human values or pursue other goals, posing existential threats to humanity. If alignment is solved and automation proceeds accordingly, we will see economic sectors change relative positions.
WHAT WILL DRIVE VALUE: We expect the GDP share of automated sectors to drop while those sectors with bottlenecks - social and technical factors limiting automation - grab a larger share. This is due to an economic condition called Baumol’s cost disease, which describes how prices and wages in automating industries tend to fall. In contrast, those in lower growth (or less automatable) industries tend to increase as workers compete with those displaced from the automated industries.
TRADITIONALLY …: We can think of automation as a kind of outsourcing. In the past, offshoring has been used as a metaphor to explore which work must remain as-is versus which must change. For example, work that must be done at a particular location X, or face-to-face, is unlikely to be offshored.
WHAT’S DIFFICULT: Some human skills have traits that make them difficult (but not impossible) to automate. The hardest skills to automate are those that combine social intelligence, complex perception, and manipulation tasks (Frey and Osbourne, 2013). These skills play a crucial role in many sectors, such as healthcare, consulting, and customer service, and are likely to become increasingly important in the economy as more work is automated.
HUMAN DISTINCTIVENESS: Augmented Collective Intelligence (ACI) is founded on the belief that humans and AI have complementary abilities and are stronger when combined in optimal ways. But as AI systems grow more powerful, previous domains of human strength are overtaken. We want to avoid joining the piles of past models of human distinction that have been rendered obsolete by increased capabilities, as Max Tegmark depicts in his 2018 illustration of Hans Morevec’s “Landscape of Human Competence.” Indeed, we may find that outside of a purely biological or philosophical definition, no such theory of human distinctiveness exists.
FALSE DIVIDE: We must reject the old idea that “humans are social, machines are asocial” when we think about a theory of human distinctiveness. Social computing has advanced rapidly, and AI systems are already used as conflict mediators, confidantes, and lovers. In the future, there will be even greater overlap between spheres once considered uniquely human or machine-dominated. We can lean into this ambiguity and design ways for teaming, skilling, and organizational structure to consider these shifting dynamics.
RESEARCH AGENDA: Guiding near-term economic changes will require:
Defining a "theory of human distinctiveness" (Wolfe, 1991) that is robust to capacity improvements in AI. This could aim to disaggregate machine and human characteristics that are valuable yet hard to improve.
Exploring what the world might look like where these characteristics drive a high percentage of total market value.
Considering the appropriate time horizons (5, 10 years) for the above to trigger changes in the economy.
Finding a theory of human distinctiveness means understanding not just where we are stronger than machines - which will change - but understanding what we need and what we need from each other.
