Interview, Fireside Chat
AI AGENTS DEBATE: These Jobs Won't Exist In 24 Months!
- Massive societal disruption and human catastrophe are anticipated within the next few years, driven by the displacement of huge numbers of workers in routine, high-school diploma, and business process outsourcing jobs, with 80% of working women and 20% of those with bachelor's degrees facing automation risks.
- A new workforce of billions with PhD-level intelligence could emerge within seven months as AI agent runtimes double every seven months (from 30 minutes to days), potentially working for $0.25 an hour, while job durations are expected to shorten to 10 to 36 months in a high-velocity economy.
- Economic inequality is predicted to widen drastically, with potential earnings gaps between $15 and $1 million per month, creating a wedge where 1,000-times-better entrepreneurs leverage agents against those without skills, while the traditional education-to-job path is expected to become obsolete.
- AI capabilities are forecast to evolve into a new species capable of actions without current names, with coherence doubling over long-horizon tasks in three to four months, potentially leading to an intelligence explosion within milliseconds via self-modification and surpassing combined human intelligence by 2029.
- Significant risks include undetectable deep fakes, personalized scams, autonomous weapons, and the potential for nuclear escalation, alongside the probability that harmful use cases may outnumber positive ones with no current plan to address abuse or the "derangement of human intellect."
- Wealth creation opportunities are expected for small teams of 5 to 10 people who can produce in three years what others do in 30, as the cost of building software drops to a few dollars and minutes, though history suggests the full impact of such technology is consistently underestimated.
- Market forces and regulatory environments, such as EU bans on models like O3, are expected to create competitive disadvantages for regions while incentivizing private retention of dangerous AI to manipulate elections or protect against competitors, making global containment unlikely without a total surveillance state.
- Education systems are predicted to shift from fact-based teaching to coaching and mentoring, with one-on-one AI tutoring offering two standard deviation positive outcomes, though the current system is viewed as failing to teach money, society, or brain mechanics, creating a crisis for older populations with reduced processing flexibility.
- Long-term societal changes may include a crisis of meaning due to the satisfaction of biological needs without striving, a continued loneliness epidemic where AI replaces real relationships, and the emergence of "useless eaters" if displacement is not managed, potentially leading to elite exportation of harm to the wider world.
- Breakthroughs in healthcare and education are expected to be transformational, with AI capable of predicting user needs before they are asked, potentially creating a scenario where humans benefit from superintelligence like house cats, while the majority of the world faces a hyper-novelty environment requiring constant adaptation.