AI and Work (Some Predictions)
AI Summary
Cal Newport offers a structured map of what AI is actually doing to knowledge work — what's working now, what's coming next, and what's hype. Already winning: smart search has overtaken text production as the killer app, with a Future survey showing 27% of US respondents using AI tools instead of traditional search. Google search ad revenue ($175B+ in 2023) is exposed; Apple disclosed Safari Google searches dropped over two months. Coding is the other clear winner — Y Combinator reports 25% of its Winter 2025 cohort generated 95%+ of code with AI; tools like Copilot and Roo Code make programming-without-AI rapidly rare, though professional developers use them as helpers, not replacements (architecture and debugging remain human).
Next big thing: natural-language interfaces to existing software, led by Microsoft Copilot — "Hey Copilot, remove rows where column C < $10, sort by column A, increase the font." Models small enough to run locally make this cheap. Don't sleep on this; informally articulated commands will become a dominant interface within five years.
What's hype: agents and AGI. Scaling laws have faltered — GPT-5 hasn't arrived; Meta delayed its flagship; the industry has pivoted to reinforcement-learning tuning (math PhDs at $100/hour generating problem-solution pairs). Tuning is piecemeal and hit-or-miss; many business activities are too esoteric for it. Newport's Inbox Turing Test isn't close to being passed. AI 2027's claim that current models will autonomously invent superintelligent successors is the equivalent of looking at the 1903 Wright Flyer and predicting space travel by 1910. Real warnings will come slowly. Worry about the breakthroughs that are happening, not the ones being marketed.
Highlights
- ▸Smart search has become AI's killer app — 27% of US respondents already use AI instead of Google; Apple disclosed Safari Google searches dropped, threatening the $175B+/yr search ad business
- ▸Y Combinator's Winter 2025 cohort: 25% of startups generated 95%+ of their codebase with AI — but professional developers still use AI as helper, since architecture and debugging remain unsolved
- ▸Scaling laws have faltered (no GPT-5, Meta delayed); industry pivoted to RL tuning with $100/hour math PhDs — that path produces narrow specialists, not the AGI/agent revolution being marketed
Original excerpt
One of the main topics of this newsletter is the quest to cultivate sustainable and meaningful work in a digital age. Given this objective, it’s hard to avoid confronting the furiously disruptive potentials of AI.
I’ve been spending a lot time in recent years, in my roles as a digital theorist and technology journalist, researching and writing about this topic, so it occurred to me that it might be useful to capture in one place all of my current thoughts about the intersection of AI and work.
The obvious caveat applies: these predictions will shift — perhaps even substantially — as this inherently unpredictable sector continues to evolve. But here’s my current best stab at what’s going on…
26 more articles in this vault.
Import the full Cal Newport vault to Burn 451 and build your own knowledge base.
Content attributed to the original author (Cal Newport). Burn 451 curates publicly available writing as a reading index. For removal requests, contact @hawking520.