Has AI Changed Work Forever? Not Really... | Cal Newport

Name: Has AI Changed Work Forever? Not Really... | Cal Newport
Uploaded: 2026-02-26T11:00:00Z
Channel: Cal Newport

February 26, 2026 | 4 products mentioned

Cal Newport host

Watch on YouTube artificial intelligence ai hype code generation machine learning progress technology criticism workplace automation ai skepticism

Cal Newport critiques a viral essay by AI entrepreneur Matt Schumer that claims AI has reached an inflection point that will fundamentally transform work. Newport argues the essay uses emotional manipulation and misrepresents the actual state of AI progress, which has slowed rather than accelerated since 2025, with meaningful improvements confined to narrow applications like code generation rather than the broader workplace transformation Schumer describes.

Key takeaways

• AI progress slowed in 2025 after pre-training scaling hit diminishing returns, forcing companies to shift toward post-training fine-tuning and niche optimization rather than the accelerating exponential gains Schumer claims.
• The essay conflates incremental improvements in code generation with transformative AI capability, when in reality only a small fraction of professional programmers use these tools without extensive human supervision and testing.
• AI coding agents cannot recursively self-improve or build better AI models—they excel at automating tedious, repetitive coding tasks like API integration and UI building, not at mathematical or conceptual innovations that drive actual AI breakthroughs.
• Based on 250+ case studies of real programmer workflows, Newport found that no one is using these tools as described in the essay (describe task, walk away for 4 hours, finished product); instead, developers maintain heavy supervision, run extensive testing, and abandon the AI ~20% of the time.
• AI companies focused on code generation not as a strategic path to self-improving systems, but because it was the only niche market where they found traction after failing to deliver on broader promises in video generation, general-purpose agents, and other domains.
• The essay represents "vibes-based thinking" common to alarmist reporting that mixes factual observations with unfounded speculation, rather than grounded analysis of where AI actually delivers value today.