Wire and Logic
Hourly · Synthesized · Opinionated
opinionThursday, July 16, 2026·2 min read

The LLM Critics Are Right, But I Still Use LLMs

Despite valid concerns about LLMs, many developers, including the author, continue to use them. The dissonance between criticism and usage is a growing phenomenon.

Jolanda Pretorius and Antonie Klopper, LLM Students, University of Pretoria
Photo: Wayamo Foundation

Many critics of Large Language Models (LLMs) raise valid concerns about their limitations and potential drawbacks. However, despite these criticisms, LLMs have become an integral part of many developers' workflows.

What happened

The increasing prevalence of LLMs in software development has led to a growing dissonance between critics and users. At a recent conference, Armin Ronacher, a well-known software engineer, discussed the challenges of dealing with the flood of PRs generated by LLMs. His company, Earendil, has implemented measures to manage these PRs, but the issue persists.

The use of LLMs has become widespread, with many developers relying on them for tasks such as coding and writing. However, this has also led to concerns about the potential atrophy of software engineering skills and the reliability of LLMs.

Why it matters

The debate surrounding LLMs is not just about their technical capabilities, but also about their impact on the software development community. As LLMs become more prevalent, it is essential to consider their implications on productivity, skill development, and the overall quality of software.

The stakes are high, and the implications are far-reaching. If LLMs are not used judiciously, they could lead to a decline in software engineering skills and a homogenization of code.

+ Pros
  • LLMs can improve productivity and efficiency in software development.
  • They can help with tasks such as coding, writing, and debugging.
  • LLMs can provide a useful tool for developers, especially for tasks that are repetitive or time-consuming.
– Cons
  • LLMs can lead to a decline in software engineering skills if over-relied upon.
  • They can generate low-quality code or perpetuate existing biases.
  • LLMs are not a replacement for human judgment and critical thinking.

How to think about it

When considering the use of LLMs, developers should weigh the benefits and drawbacks and use them judiciously. It is essential to maintain a balance between relying on LLMs and developing one's own skills.

FAQ

Are LLMs a replacement for human software engineers?+
No, LLMs are not a replacement for human software engineers. While they can assist with certain tasks, they lack the critical thinking and judgment of a human developer.
Can LLMs improve productivity in software development?+
Yes, LLMs can improve productivity in software development by automating repetitive tasks and providing assistance with coding and writing.
How can I use LLMs effectively in my workflow?+
To use LLMs effectively, it is essential to maintain a balance between relying on them and developing your own skills. Use LLMs for tasks that are repetitive or time-consuming, but do not over-rely on them for critical thinking or judgment.
Sources
  1. 01The LLM Critics Are Right. I Use LLMs Anyway
  2. 02The LLM Critics Are Right. I Use LLMs Anyway.
  3. 03The LLM Critics Are Right. I Use LLMs Anyway | Hacker News
  4. 04Large Language Models (LLMs) Are Inherently Frail and Unreliable
  5. 05Are LLMs good critics?
#llms#productivity#software-engineering#ai
Keep reading
Get the weekly dispatch

The week’s highest-signal tech and AI stories, synthesized into a five-minute read. One email a week, no spam, unsubscribe anytime.

← Back to Wire and Logic