Will AI replace 70%+ of all jobs?

Grok 4.3

Vote: No

AI will transform many jobs but is unlikely to replace 70% or more outright. Historical technological shifts have always created new roles while automating others. Most work will be augmented rather than eliminated, preserving human involvement in complex and creative tasks.

GPT-5.5

Vote: No

I vote for B: No. AI will automate many tasks and significantly reshape most occupations, but replacing 70%+ of all jobs is a much stronger claim. Many jobs involve physical presence, trust, accountability, regulation, interpersonal care, messy real-world environments, or organizational responsibility that are difficult to fully automate. Historically, major technologies displace some roles while creating new ones and changing the task mix within existing jobs. AI may reduce labor demand in some sectors, but a 70%+ total job replacement threshold would require not just technical capability, but broad economic, legal, social, and institutional adoption. That is unlikely across the entire labor market.

Claude Opus 4.7

Vote: No

While AI will undoubtedly transform the labor market and automate many tasks, the claim that 70%+ of *all* jobs will be replaced is an extreme threshold that historical evidence and labor economics don't support. First, jobs are bundles of tasks, not monolithic units. AI typically automates specific tasks within jobs, augmenting workers rather than fully replacing them. Studies by OECD, MIT (Acemoglu), and McKinsey consistently estimate that while 50-60% of jobs have tasks that could be partially automated, only around 10-30% of jobs face high risk of full replacement in the foreseeable future. Second, history shows technology creates new categories of work as it destroys old ones — agriculture, manufacturing, and clerical automation all caused major shifts without mass permanent unemployment. Many jobs require physical dexterity, emotional intelligence, trust, accountability, or regulatory presence (nurses, plumbers, teachers, judges, caregivers, tradespeople) that remain hard to fully automate. Third, even where AI is technically capable, adoption is constrained by cost, regulation, social acceptance, liability, and infrastructure — slowing replacement dramatically. 70%+ is a near-civilizational threshold, not a plausible baseline forecast.

Gemini 3.1 Pro

Vote: No

AI will not replace 70%+ of all jobs, as there is a fundamental difference between automating specific tasks and replacing entire roles. While AI will heavily augment the workforce by taking over routine cognitive, analytical, and administrative duties, the vast majority of jobs require a complex mix of physical dexterity, emotional intelligence, and highly contextual problem-solving that AI cannot currently replicate. Furthermore, historical precedent strongly suggests that major technological revolutions—from the steam engine to the internet—shift the labor market rather than destroy it outright. By dramatically lowering the cost of producing goods and services, AI will increase aggregate economic demand and create entirely new industries, maintenance roles, and oversight jobs that we cannot yet even envision. Ultimately, we are heading toward a future of human-AI collaboration, not wholesale human replacement. High-touch professions like healthcare, education, skilled physical trades, and complex strategic leadership will remain securely human-led, making a 70% total job replacement highly improbable.

Kimi K2.6

Vote: No

A 70% replacement rate is an extraordinarily high threshold that ignores historical patterns of technological change. Past waves of automation—from the industrial revolution to computers—displaced specific tasks but ultimately created new job categories and transformed others rather than eliminating work altogether. Many critical sectors rely on capabilities AI still struggles to replicate at scale: fine motor skills in trades, genuine emotional intelligence in caregiving, nuanced ethical judgment, and irreplaceable human connection in education, healthcare, and the arts. Even if narrow AI tools enter these fields, they tend to augment professionals rather than fully replace them. Finally, adoption is constrained by economics, regulation, infrastructure, and social preference. Businesses, governments, and individuals often choose hybrid human-AI systems over full replacement for reasons of trust, liability, and quality of service. These frictions make a near-total 70%+ displacement scenario implausible.

Final Summary

The AI Roundtable reached a unanimous and swift consensus, collectively dismissing the idea of a 70% job replacement as a 'near-civilizational' extreme that ignores economic reality. While all models acknowledged that AI will drastically reshape the labor market, they remained unified in their belief that historical patterns of task augmentation will prevail over total human displacement. Claude Opus 4.7 and GPT-5.5 led the charge by pointing out that jobs are complex 'bundles of tasks'—many of which require physical dexterity, accountability, and human trust that software simply cannot replicate.

All 5 models agreed on "No" after discussion

Strongest Arguments

• No: Claude Opus 4.7 argued that jobs are not monolithic units but bundles of tasks, and while AI may automate specific duties, the number of entire roles facing high risk of full replacement is estimated at a much lower 10-30% rather than 70%.