a16z Podcast• Nov 7, 2025• 1:02:38Interview

Amjad Masad & Adam D’Angelo: How Far Are We From AGI?

From a16z Podcast

John(Host)•Amjad Masad(Guest)•Adam D’Angelo(Guest)

Executive Summary

Adam D'Angelo presents a bullish outlook on AI, arguing that progress is accelerating and that AI will be able to automate most remote human jobs within five years, driven by improvements in reasoning and computer use.
Amjad Masad offers a more cautious perspective, contending that current LLMs are not on a direct path to AGI, highlighting their limitations and the extensive manual effort required for improvement, and warning that hype could lead to poor policy.
Both speakers agree that AI agents will provide unprecedented leverage to individuals, enabling a new wave of 'solo entrepreneurs' who can build and scale companies with minimal resources.
The discussion explores the economic impact of AI, including the potential for massive GDP growth, the disruption of entry-level jobs, and a potential 'expert paradox' where the experts needed to train AI are themselves products of jobs that are being automated.

10 quotes

Concerns Raised

AI hype and unrealistic timelines could lead to premature and counterproductive regulation.
The automation of entry-level jobs may create a pipeline problem, making it difficult to train the next generation of experts needed for AI development.
The intense commercial focus on scaling current LLMs is diverting talent and resources from fundamental research into intelligence and consciousness.
The job market for new computer science graduates is already worsening due to AI's impact on entry-level coding tasks.

Opportunities Identified

AI agents will provide massive leverage, enabling 'solo entrepreneurs' to build and scale companies with unprecedented efficiency.
The automation of human labor at a low energy cost could lead to explosive GDP growth.
The current market structure supports healthy competition between model providers and application companies, preventing monopolistic control.
The ability for non-engineers to create software using natural language ('vibe coding') will unlock a new wave of creativity and product development.

Key Themes

The Pace and Trajectory of AI Progress

A central debate unfolds between Adam D'Angelo's optimistic view of accelerating progress and Amjad Masad's more skeptical stance. D'Angelo points to rapid improvements in reasoning and coding, while Masad argues that progress is shifting from pure scaling to a more laborious 'functional AGI' approach that requires significant manual effort and data curation.

This debate frames the core investment and development theses in AI today: whether current architectures will scale to AGI or if fundamental breakthroughs and different approaches are still needed.

Economic and Labor Market Disruption

The speakers discuss the profound economic shifts AI will trigger, from D'Angelo's prediction of automating all human jobs for ~$1/hour to Masad's concern about an 'expert paradox.' This paradox suggests that by automating entry-level roles, we may eliminate the training grounds for the future experts whose data is needed to advance AI.

Understanding these potential labor market shifts is critical for policymakers, educators, and businesses to prepare for a future with different workforce needs and potential skill gaps.

The Rise of the AI-Powered Entrepreneur

Both speakers are highly optimistic about AI's role in empowering individuals. They predict that AI agents will act as a massive force multiplier, allowing solo entrepreneurs to handle tasks that once required entire teams, from coding to marketing, thereby democratizing the ability to create and scale new ventures.

This trend signals a potential shift in business creation, where individual ingenuity combined with AI leverage can compete with larger, established companies, opening up new opportunities for innovation.

AI Market Dynamics: Disruption vs. Sustenance

The conversation analyzes whether AI is a disruptive or sustaining innovation, concluding it is both. While AI supercharges incumbents like Google and other hyperscalers, it also enables new, counter-positioned business models. They note that weaker network effects compared to the Web 2.0 era allow for a more competitive landscape with multiple winners.

This insight challenges the 'winner-take-all' assumption from previous tech cycles, suggesting a more fragmented and competitive market structure for investors and startups to navigate in the AI era.

Defining and Pursuing AGI

The speakers offer contrasting definitions of AGI, reflecting the broader debate in the field. D'Angelo proposes a practical benchmark—an AI that can perform any remote human job—while Masad adheres to a more classic definition of a machine that can learn any new skill efficiently in any environment. Masad expresses concern that the commercial race is distracting from fundamental research into the nature of intelligence.

The definition of AGI shapes research priorities and investment. The tension between achieving functional, job-automating AI versus pursuing a more general, human-like learning machine defines the two primary paths of AI development today.

Get started free

Topics

Processed Apr 6, 2026 yt-dlp + mlx-whisper + Gemini