Forget UBI. You’ll Get Paid Just for Being Human

Summary

Tom Bilyeu interviews mathematician and AI entrepreneur Emad Mostaque about how rapidly improving AI could reshape work, ownership, and society. Mostaque argues that AI’s economic impact may come less from a sudden arrival of “superintelligence” than from increasingly competent, reliable, and inexpensive systems that can be integrated into everyday workflows. His timelines and forecasts are predictions, not settled outcomes.

AI capabilities and technical developments

• More reliable AI: Mostaque says newer models are becoming better at avoiding mistakes and producing useful work, moving beyond the conspicuous errors and awkward output of earlier systems. He points to improvements in mathematics, coding, design, video generation, and forecasting.
• Math and formal verification: The discussion covers AI systems tackling difficult mathematical problems, including parts of a Navier–Stokes problem. Mostaque describes using large teams of AI agents and formal proof-checking tools, while noting that the work still requires review and acceptance. He also cites AI-generated proofs he considers novel, though the transcript does not establish their independent validation.
• Agent teams: The speakers distinguish ordinary chatbots from agents that can take actions, such as booking a restaurant or checking schedules, and from swarms of agents working together. Mostaque argues that teams of agents could coordinate at scales and speeds humans cannot match.
• Multimodal generation: They discuss improvements in video models’ ability to represent physical behavior, along with AI-generated 3D scenes, websites, and music. The claimed advance is not just more polished output, but models that appear to grasp concepts such as object behavior and spatial relationships.
• Lower inference costs: Mostaque says model improvements, software “harnesses” that guide how a model works, and more specialized hardware could make comparable AI work dramatically cheaper and faster. A harness is code and rules around a model that help it complete longer tasks and improve its approach. He also discusses compressing model weights so smaller models can run on less memory, and chips that store model weights directly on silicon to speed up inference.
• Economic bottlenecks: As AI becomes faster and cheaper, Mostaque expects people—not computing speed—to become a limiting factor in many information-processing workflows.

Jobs, ownership, and economic forecasts

Mostaque predicts that digital and other cognitive work will face pressure as AI becomes reliable enough to handle routine tasks. He forecasts substantial white-collar job losses, while acknowledging that adoption, integration, and the availability of physical robots will affect the pace.

The speakers discuss how companies may reduce hiring without laying off everyone at once: AI could allow smaller teams to produce more, meaning fewer new roles are created. Mostaque is skeptical that cheaper AI will ultimately create enough new jobs to offset displacement.

He expects a widening divide between people who use AI effectively, people who own AI infrastructure such as chips and data centers, and people who have neither. He also sees robotics as a later but significant source of disruption, while recognizing manufacturing constraints.

As a near-term response, Mostaque advises people to use AI regularly and examine their jobs task by task:

1. Identify each task’s input and output.
2. Test whether AI can perform it adequately.
3. Use the results to automate routine work or focus on work that remains valuable.

He also suggests that AI-enhanced entrepreneurship could help people build businesses with smaller teams.

The title’s idea of being “paid just for being human” is presented as a possible future alternative to conventional tax-funded universal basic income (UBI). Mostaque argues that the tax base may not support a broad UBI if employment and the cost of intelligence change dramatically. He speculates that future income or support could instead come through mechanisms such as shared ownership of AI infrastructure, robots, or other productive assets, or through universal basic services. He does not present a settled plan for how such a system would work.

Intelligent Internet and AI governance

Mostaque describes Intelligent Internet, his proposed effort to build locally owned AI infrastructure and services. He compares the concept to a credit union or utility: citizens and local institutions would own a stake, potentially including shares reserved for children, and the system could provide AI agents and invest in data centers and robots.

A central proposal is an agent for each person that could coordinate access to services and support both collective and individual goals. The models and governance would need to reflect explicit, transparent commitments and values, rather than leaving consequential decisions to opaque systems owned by outside companies.

The discussion raises concerns about privacy, bias, advertising incentives, political influence, and people becoming overly dependent on AI.

Social risks and proposed responses

The speakers consider what people might do if paid work no longer provides income, identity, status, or a sense of progress. Possibilities discussed include infrastructure projects, community service, digital worlds, stronger local communities, and new ways of distributing access to scarce goods. They also consider severe possibilities—including social unrest, entrenched inequality, and debt relief—without agreeing on a single solution.

Mostaque’s closing recommendations are to learn to use AI, look for entrepreneurial and ownership opportunities, and build stronger communities. Both speakers stress that society will need to reconsider how people find purpose and how economic gains are shared.

Products and guides mentioned

• Opus Clip (sponsored segment): The ad says it can turn long videos into short clips, reframe footage for vertical formats, upscale video, generate relevant B-roll, add captions, motion graphics, and sound effects, and dub videos into other languages. This is a product promotion, not an independent review.
• Practical guide offered in the discussion: Break a job into input-to-output tasks and test whether AI can perform each one. No separate step-by-step tutorial or product review is provided.
• Other sponsored segments: AT&T Business and Ethos life insurance appear, but are not substantively reviewed in the conversation.

Main speakers and sources

• Tom Bilyeu — Host and interviewer.
• Emad Mostaque — Guest; mathematician and AI entrepreneur discussing AI capabilities, labor-market forecasts, and Intelligent Internet.
• Other sources referenced: OpenAI, Anthropic, Meta, Andrej Karpathy, and AI forecasting and mathematics projects. Their claims are conveyed through the conversation rather than independently assessed in the video.