Cal Newport AI takes are WILD...

2026-03-04 · Source: Wes Roth · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Emerging Technologies & Innovation · Depth: Intermediate, extended

Summary

This analysis critically refutes Cal Newport's commentary on AI progress, particularly his claims regarding a slowdown in 2025 and the "grade A nonsense" of AI self-improvement. The author presents counter-evidence, including METER Research and ARC EGI benchmarks, which show exponential AI capability growth, especially after the shift to inference-time compute models like GPT-4, Claude Opus 4.6, and Gemini. The discussion highlights AI's increasing autonomy in coding, app development, and even complex mathematical problem-solving, citing Google DeepMind's Alpha Evolve and Fun Search, Anthropic's internal AI development, and Sakana AI's self-improving models. The piece also challenges Newport's assertion of nervous investors and niche markets, pointing to massive infrastructure commitments by tech giants and record-breaking funding rounds for AI companies, alongside parabolic revenue growth.

Key takeaway

For AI Engineers and Data Scientists evaluating the pace and scope of AI capabilities, recognize that the shift to inference-time compute has driven exponential progress, not a slowdown. Do not dismiss AI's self-improvement potential or its broad market impact based on expert skepticism, as historical patterns show disruptive technologies are often underestimated by incumbents. Focus on integrating advanced AI agents for tasks like autonomous code generation and system optimization to capitalize on accelerating development cycles.

Key insights

AI progress, self-improvement, and market growth are accelerating, contrary to claims of slowdowns and limited applications.

Principles

• Disruptive tech adoption is often misjudged by experts.
• AI's ability to write code accelerates its own development.
• Exponential growth can appear incremental to casual observers.

Method

AI models, particularly those with inference-time compute, can autonomously generate code, solve complex problems, and even optimize their own hardware and software, leading to recursive self-improvement.

In practice

• AI agents can build functional websites from natural language prompts.
• AI can autonomously solve complex mathematical problems.
• AI can optimize data centers, chip design, and AI training processes.

Topics

• AI Progress Acceleration
• AI Self-Improvement
• Autonomous Code Generation
• AI Industry Investment
• AI Benchmarking

Best for: AI Engineer, Data Scientist, Tech Journalist

Related on AIssential

• The Perils of the AI Exponential — AI agent capabilities, measured by task time horizon, are accelerating dramatically, prompting economic disruption discussions.
• A Brain is Alive in a Computer — Current AI progress is incremental and specialized, not a general exponential acceleration or self-improving loop.
• Full automation of AI R&D probably yields a large speed up even without a software-only singularity — Full AI R&D automation will greatly accelerate progress through initial speed-ups and enhanced compute returns, even without a singularity.
• Human Amnesia Is Blinding Us to AI’s Speed — Human cognition struggles to grasp exponential change, leading to underestimation of rapid technological acceleration.
• The Race to Recursive Self-improving AI and Exponential Tech — The AI industry is rapidly pivoting from AGI to Recursive Self-Improving AI, driven by startups and significant investment.
• The Annual AI Slowdown Panic Is Here — The AI industry is navigating a resource-constrained era, shifting focus to efficient inference and realistic benchmarking amidst evolving job impact narratives.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Wes Roth.