TL;DR
Veteran programmer antirez argues that current large language models are already altering how software is produced, often completing substantial coding tasks with limited human guidance. He documents recent personal experiments, endorses open source, warns about job displacement and centralization, and urges developers to engage with AI tools rather than ignore them.
What happened
In a recent essay, Redis creator antirez described a rapid shift in software development driven by modern LLMs. After tracking AI developments and starting an AI-focused YouTube channel late in 2024, he recounts several recent experiments where prompt-driven models produced or fixed significant code in hours instead of weeks. Examples include adding UTF-8 support and test harnesses to a line-editing library, resolving intermittent Redis test failures, generating a pure C inference library for BERT-like embeddings (about 700 lines, roughly 15% slower than PyTorch), and reproducing internal Redis Streams changes from a design doc. He says these experiences convinced him that hand-writing most code is becoming unnecessary for many projects. He also expresses enthusiasm about his work being incorporated into training data, renewed focus on open-source contributions, and concern about possible job losses and AI centralization. His practical advice: test AI tools seriously and adapt rather than refuse them.
Why it matters
- AI-assisted coding can shorten project timelines from weeks to hours for many tasks, altering developer workflows.
- Smaller teams and individual maintainers could gain capabilities that previously required larger organizations.
- Widespread automation raises questions about job displacement and the need for social or political responses.
- Concentration of model development in a few firms risks reversing current degrees of AI democratization.
Key facts
- The author left his job in 2020 to write a novel about AI and related social issues.
- He launched an AI-focused YouTube channel at the end of 2024 to explore coding uses and societal effects.
- He reports recent LLMs can complete large subtasks or medium projects with good prompts and occasional human guidance.
- Four cited projects completed in hours: UTF-8 support and testing for linenoise, fixing Redis transient test failures, a pure C BERT-like embedding inference library, and reproducing Redis Streams internals changes from a design document.
- The C inference library amounted to about 700 lines and ran about 15% slower than PyTorch, according to the author.
- He feels positive about his publicly available code being used by LLMs and sees that as extending his goals of democratizing code and knowledge.
- The author endorses producing more open-source software and applying AI to ongoing Redis work.
- He worries about potential mass layoffs in programming and the political pressures that could follow if many people lose jobs.
What to watch next
- Whether companies respond to AI by expanding teams and building more, or by cutting headcount and relying on fewer prompt-skilled engineers.
- Moves by governments and voters to create social protections for those displaced by automation, as suggested by the author.
- The degree to which model development consolidates in a few labs versus remaining accessible through open and competing models.
Quick glossary
- Large language model (LLM): A machine learning model trained on large amounts of text that can generate, transform, or reason about language and code.
- Inference: The process of running a trained model to produce outputs from inputs, such as generating code or embeddings.
- Open source: Software whose source code is made available for use, modification, and distribution by anyone.
- Reinforcement learning: A training approach where models learn by receiving feedback or rewards for actions, often used to refine behavior.
Reader FAQ
Does antirez believe AI will replace programmers entirely?
He argues that for most projects, writing code yourself is often no longer sensible, but he does not claim programmers will disappear; instead he encourages adapting to new tools.
Did he test AI tools on real engineering work?
Yes — he describes using LLMs to add UTF-8 support, debug Redis test failures, create a C inference library, and reproduce Redis Streams changes.
Is antirez worried about AI centralization?
Yes — he expresses concern that the technology could become concentrated in the hands of a few companies and calls for continued openness.
Should developers ignore AI tools to protect their careers?
Not confirmed in the source
Don't fall into the anti-AI hype antirez 1 hour ago. 12018 views. I love writing software, line by line. It could be said that my career was a continuous effort…
Sources
- Don't fall into the anti-AI hype
- The great AI hype correction of 2025
- With generative AI, don't believe the hype (or the anti-hype)
- Beyond the Hype: When Generative AI Isn't Always …
Related posts
- European Commission launches open-source drive to curb reliance on US Big Tech
- PrintReadyBook: AI tool that generates print-ready books with cover art
- Sisyphus Now Lives in Oh My Claude: Multi-agent orchestration for Claude Code