TL;DR
A software consultant argues that large language models sap the pleasures of both programming and management. He uses LLMs reluctantly for client work, saying they can be effective but at the cost of craftsmanship and the joy of mentoring.
What happened
In a personal essay published on December 28, 2025, a professional consultant described how his relationship with programming and team leadership has changed since LLMs became widely available. He says he originally entered software to translate clear thought into reliable machine behavior and later found the greatest satisfaction in mentoring colleagues and watching them grow. Now, because he advises clients across product, architecture and team building, he feels obliged to use and guide others in LLMs. He describes other people's framing of these models — as the ultimate programming tool or an extra teammate — as a "dark parody." For him, using LLMs to write code is like hiring someone to solve his puzzles, and attempting to manage or nurture an LLM is either pointless or infuriating. He can operate them effectively, he writes, but doing so costs him care for his craft and removes joy from the work.
Why it matters
- Adoption of LLMs can change the day-to-day satisfaction engineers derive from problem-solving and craft.
- Tools that automate coding may alter how managers invest in mentorship and team development.
- Even when productive, LLM workflows can impose emotional and professional costs on practitioners.
- Companies that integrate LLMs may need to reconcile efficiency gains with potential impacts on team morale and skill growth.
Key facts
- Author began programming to turn precise reasoning into reliable machine behavior.
- Over time the author grew to value mentoring colleagues and watching them learn.
- The author consults for multiple companies on product direction, team building, architecture and code.
- He says he regularly uses LLMs for projects and guides clients on their effective use.
- Others reportedly call LLMs the ultimate programming tool or an extra machine teammate; the author rejects those labels.
- The author describes LLMs as a "dark parody" of those descriptions and states plainly: "LLMs are not fun."
- Using LLMs to write code is compared to hiring a taskrabbit to solve the author's jigsaw puzzles.
- Nurturing an LLM is depicted as a waste of time, and micromanaging them as frustrating and rage-inducing.
- The author can make effective use of LLMs but says doing so costs his care for his craft and the joy the work used to bring.
What to watch next
- How teams balance LLM-driven productivity with opportunities for human mentorship and skill growth — not confirmed in the source.
- Whether organizations change management practices to address the emotional and craft-related costs described — not confirmed in the source.
- Trends in developer job satisfaction after widespread LLM integration — not confirmed in the source.
Quick glossary
- LLM: Large language model: a type of AI trained on large text corpora to generate or predict human-like text.
- Mentorship: The process of experienced practitioners guiding less experienced colleagues to help them develop skills and judgement.
- Micromanagement: Close, detailed control or supervision of tasks and people, often viewed as counterproductive to autonomy and growth.
- Craftsmanship (in software): Attention to quality, clarity, and deep understanding of a system when designing and writing code.
Reader FAQ
Does the author still use LLMs?
Yes. The author says he regularly uses LLMs for projects and advises clients on their use.
Does the author find LLMs enjoyable?
No. He states that "LLMs are not fun" and argues they undercut the joys of programming and management.
Does the author believe LLMs are the ultimate programming tool or a team member?
He notes others describe them that way but calls those descriptions a "dark parody."
Will LLMs replace human mentorship?
Not confirmed in the source.
orib.dev: LLMS Are Not Fun I got into programming because it let me turn thought into reality. If I could reason precisely, then a computer could carry out my thoughts…
Sources
- LLMs Are Not Fun
- Why SWEs Love LLMs, But “Coders” Are Freaking Out
- Firing programmers for AI is a mistake
- The Disappearing Middle: How AI Coding Tools Are …
Related posts
- 2025: When AI’s Frenzy Met a Hard Reality Check on Growth
- Why the U.S. electrical grid increasingly relies on software solutions
- Using ChatGPT’s new app integrations: DoorDash, Spotify, Uber and more