We cover AI, so we owe you unusual clarity about how we use it ourselves.
The short version
Humans write our articles. We do not publish AI-generated articles under a human name, and we do not publish AI-generated articles at all.
What we do use AI for
The same things anyone uses software for. Specifically:
- Transcribing recorded interviews and calls
- Summarising long documents we then read ourselves
- Copy checking, spelling and grammar
- Searching and retrieving sources, which we then open and verify
- Generating test inputs when we are benchmarking an AI product
What we do not use AI for
- Writing or drafting articles, reviews or newsletters
- Producing verdicts, scores or recommendations
- Generating quotes, statistics or sources
- Writing anything published under a person’s byline that they did not write
- Creating images presented as photographs of real products or people
AI-generated imagery
Where we use an AI-generated illustration, the caption says so. We never use AI-generated imagery to depict a real product interface, a real event, or a real person.
Why this matters for a review site
An AI system can produce a fluent paragraph about a product it has never run. That is precisely the failure mode we exist to avoid. If a review could have been written without opening the product, it has no value, regardless of who or what wrote it.
Testing AI products with AI tools
When we benchmark an AI product, we sometimes use other AI systems to generate test cases or to grade outputs at scale. Where we do this, the review says so, describes the setup, and reports the human spot-check rate, because automated grading has its own failure modes.
Holding us to it
If something we published reads like machine output, tell us: contact@uncommonbits.com. We would rather hear it from you than from a reader who quietly stops trusting us.
Last updated 6 September 2026.