Reading up on LLMs

Terry Tao contends that LLM-produced mathematical results are a source of pride, representing the collective achievement of the mathematical community rather than a threat to it.

LLMs' strength in math and coding comes from the abundance of relevant pretraining data they see, not from the ease of verifying answers.

AI coding agents lower writing costs but shift the burden to understanding generated code, arguing that human comprehension must remain a core design goal.

Twenty‑five Fields Medalists warn that AI companies’ focus on solving math problems undermines deep understanding and risks misaligning AI progress with the goals of mathematics.

Shopify is moving its major mobile apps from React Native to native Swift and Kotlin, using LLMs to cut the cost of building and maintaining features twice.

Engineering management must refocus on auditing assumptions, measuring business outcomes, and owning specification and accountability, as AI-driven code production makes traditional output metrics misleading.

Hacker News commenters express nostalgic love for computers, lament modern AI non-determinism, yet acknowledge LLMs as useful for rapid prototyping.

LLMs often create the illusion of productivity while actually reducing real output through cognitive debt, maintenance overhead, and misaligned incentives.

Software teams costing €130k per engineer per year must generate 3–5× that to be viable, yet most lack financial visibility, a condition now exposed by AI.

Because LLMs drastically lower the cost of cleaning up messy code, engineers should take more intentional technical debt now and trust future models to service it.

Falling costs of LLM-generated code shift software engineering bottlenecks to infrastructure, CI, human-guided abstractions, and code review in 2026.