How we use AI

Where AI fits into our testing — and where it doesn't

We test AI calorie trackers, and we also use AI tools in our own workflow. Here is exactly where, and the limits we put on it.

Where we use AI

• Data processing. We use scripts and models to help organize logged results and flag outliers in our accuracy data for human review.
• Drafting support. AI may help draft or tidy explanatory passages, which a named editor then rewrites, fact-checks, and signs.
• Search and summarization. We use AI to surface app changelogs and feature changes faster, which we then verify by hand in the app.

Where we never use AI

• Scoring. Benchmark scores come from measured results against our dataset, not from a model's opinion.
• Final judgment. Rankings and recommendations are decided by named humans, including a registered dietitian.
• Medical claims. Health, weight, and metabolic statements are reviewed by a clinician, never published on a model's say-so.

Our standard

AI can make our research faster, but it cannot be the reason you trust a number on this site. Every figure traces back to a documented test and a human who stands behind it. See our methodology and editorial policy.