Google DeepMind Asks What Work Means in the AI Era

What Happened

Google’s DeepMind Institute published two essays examining how artificial general intelligence or AGI, could transform work. AGI is AI capable of a wide range of human-level tasks. One essay studies how AI systems might compete with human teams. The other asks what people gain from employment beyond money.

The Economics of Intelligence: How and When Does AGI Lead to Economic Transformation? This essay focuses on AI agents, which can complete multiple steps autonomously to finish tasks. Its central distinction is between an agent’s individual skills and the ability of multiple agents to divide work and share information. The researchers say today’s agents still coordinate poorly and offer no timetable for economic transformation.

What Is Work For? Human Wellbeing and the Value of Work in the AGI Era examines purpose, social connection, status, and agency. Its survey found that most workers would leave their jobs if salary and benefits were guaranteed, but a substantial minority would stay even with more money offered to leave.

SmarterX founder and CEO Paul Roetzer explored what these questions mean for business leaders, on Episode 249 of The Artificial Intelligence Show.

The Key Numbers

8,000+ - Working-age adults in the work-and-wellbeing survey

4  - Countries the survey covered (U.S., U.K., Denmark, and India)

1 in 5 - Workers willing to stay despite an offer of up to 120% of their pay to leave

165 - Work-activity categories in OpenAI’s separate ChatGPT usage dataset

6  - Occupations initially covered in Mercor’s TaskNet research

Why AI’s Impact Depends on How Work Is Organized

Teamwork matters alongside individual ability. A strong human coder and a strong writer could together outperform agents that each beat the average person at both tasks, since copies of the same AI may share weaknesses. If agents learn to coordinate, businesses could assemble much larger AI teams without the communication limits that constrain human organizations.

“We are now talking about what's the impact of agents that can do long horizon tasks on the economy,” Roetzer says. That changes the question from how much time one employee saves to how a company organizes entire workflows.

Experienced people could become more valuable as coordinators. Roetzer describes effective collaboration simply: “They're able to work across areas of expertise.” In the model’s possible scenarios, people whose skills AI can match could face pressure on wages, while specialists and managers with complementary abilities could gain.

“We might not see massive unemployment, but you're going to see leaner startups, flatter org charts, softer entry-level hiring, and then the steady rise for experienced people, business people, not just engineers, who can direct this work across systems.”

— Paul Roetzer, founder and CEO of SmarterX, Episode 249 of The Artificial Intelligence Show

SmarterX Take

Leaders need to consider what employees retain when tasks move to AI. The wellbeing essay acknowledges that many jobs are unpleasant and distinguishes involuntary unemployment, associated with worse wellbeing, from chosen retirement, which often fares better. People’s preferences, relationships, and sense of agency deserve attention alongside productivity.

That makes job redesign a human question as well as an operational one. Even employees who remain employed may lose meaningful parts of their work. The researchers suggest caregiving, volunteering, and other activities could provide some benefits of employment, but replacing income alone may not preserve everything a job provides.

What to Watch

Coordination will be a revealing measure of AI progress. Follow whether agents can reliably work across tasks and teams, and how that changes wages, hiring, and management responsibilities. Individual test scores capture only part of that picture.

Emergence of better maps of everyday work. OpenAI’s usage data and TaskNet’s worker interviews offer different views of tasks and AI use. They can help researchers study change without treating a job title as a complete description of someone’s work.

How Professionals Expect AI to Change Jobs

Seventy-one percent of respondents to the 2026 State of AI for Business Report believe AI will eliminate more jobs than it creates over the next three years. That measures expectations, not a forecast. It gives leaders useful context for conversations about changing roles, even while the pace and shape of disruption remain uncertain.

Based on more than 2,100 responses from professionals across roles, functions, and industries, the report examines adoption, organizational readiness, and workforce concerns. Its findings can help leaders understand what their people may be anticipating as AI takes on more work. Read the full report →