# Linus Torvalds Shares Critical Views on AI and Code Generation

Torvalds emphasizes 'mindless patches' from LLMs and the crucial role of human oversight.

- **Source:** [Slashdot](https://brifing.si/source/0f34d58e-1ccd-45a6-ab82-2de4065dc7ad)
- **Published:** 2026-07-12
- **Topic:** [Data & AI](https://brifing.si/c/data-ai)
- **Original article:** <https://linux.slashdot.org/story/26/07/12/2053201/linus-torvalds-on-ai-junk-patches-humans-and-godzilla?utm_source=rss1.0mainlinkanon&utm_medium=feed>
- **On Brifing:** https://brifing.si/s/1d2b4158-fbe8-4f87-9671-f051555bc8be

## Generative AI's Impact on Programmer Productivity

Linus Torvalds dismissed his previous claim about LLMs potentially increasing programmer productivity 10X as 'not scientific.' Current observations show these technologies currently produce more junk code than useful output. This imbalance was particularly noticeable in early 2026.

## The Critical Role of Human Oversight

Torvalds stresses that bug reports generated by LLMs alone aren't sufficient. Effective use requires active developer participation: 'Finding a bug and having an LLM write a report isn't enough. Humans need to propose solutions and engage in the process.'

## Dangers of Superficial Fixes

Torvalds describes most AI-generated patches as 'mindless band-aids' that only offer temporary solutions. The core issue often resurfaces in different forms. LLMs prove particularly inadequate for kernel-level fixes according to his experience.

## Early Detection of Security Vulnerabilities

Torvalds calls some security vulnerabilities detected by AI 'painfully interesting.' While their media exposure days later proves embarrassing, he emphasizes early detection over 'shooting the messenger.' He particularly values finding critical bugs that went unnoticed for 20 years.

## Using LLMs as Prototyping Tools

For personal projects, Torvalds uses LLMs as prototyping tools, acknowledging the generated code usually isn't directly usable but valuable for testing ideas. He also mentions fun applications like sending Godzilla photos to his kids during his India trip.

## Balanced Assessment of AI

Torvalds objectively evaluates both the benefits and limitations of AI. While recognizing its potential, he emphasizes it can't replace human expertise in software development. His 'Godzilla' metaphor highlights the importance of choosing appropriate applications for the technology.

## Sources covering this story

- https://linux.slashdot.org/story/26/07/12/2053201/linus-torvalds-on-ai-junk-patches-humans-and-godzilla?utm_source=rss1.0mainlinkanon&utm_medium=feed — 

---

_Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._
