How We Use AI

Last updated April 21, 2026

This document outlines the current policy regarding the use of AI tools at Independent Reviews. We make this information available as we cover the calorie tracking application category, which includes AI-driven photo recognition apps. It would be inconsistent to leave our own use of AI unrecorded while we evaluate others. To summarize, we employ AI tools as aids for research and editing, rather than as authors. Each page on this website is crafted and approved by a named human contributor.

What we use AI for

We utilize AI tools (mainly Anthropic Claude and OpenAI ChatGPT) for three types of tasks, and we restrict their use to these:

1. Research summarization. If a contributor needs to review a 60-page validation study, an AI tool might generate a preliminary summary that the contributor then compares to the original PDF. The AI summary serves as a navigation tool; it is never considered the cited source. All claims in published articles originate from primary sources that the contributor has examined directly.
2. Citation finding. If a contributor recognizes that a claim requires supporting literature but lacks a specific citation, an AI tool may be utilized to identify potential citations. Each suggested citation is then independently verified by the contributor; they will open the paper, examine the methods, ensure the claim is substantiated, and confirm the publication details. Citations suggested by AI that cannot be verified independently are discarded. We have identified fabricated citations from AI tools and view the verification rate as a quality measure for contributors.
3. Copy editing. An AI tool may assist in proposing line-level prose enhancements (for clarity, sentence flow, and conciseness) on a draft that the contributor has previously composed. The contributor reviews each suggestion and decides whether to accept or reject it. AI tools are not utilized to develop outlines into prose, create new paragraphs of content, or rewrite drafts in someone else's voice.

What we do not use AI for

The following applications are not allowed under our editorial policy:

• Primary writing. No paragraph on Independent Reviews has been generated by an AI tool. Every word published here originates from a named human contributor.
• Score generation. No app score, MAPE value, or ranking position on this site has been produced by an AI tool. Scores are derived from primary testing conducted by Sebastian and Jonah; statistics are calculated by Mei-Lin based on raw data; rankings are based on the published rubric, not an LLM evaluation.
• Test-protocol substitution. We do not request an AI tool to estimate the calorie content of a meal instead of weighing it. We do not ask an AI tool to "predict what an app would log" instead of downloading the app and recording the meal.
• Synthetic citations. We never release a citation that we have not independently verified, regardless of its source.
• Contributor impersonation. We do not engage in ghostwriting a published byline using AI under the contributor's name. The byline indicates who authored the piece.

Why we draw the line where we do

There are three primary reasons. Firstly, AI tools can generate fictitious citations and facts at significant rates. The repercussions of publishing a fabricated citation in content related to clinical claims (such as protein targets, GLP-1 nutrition, body-recomposition framing) are substantial; the advantage of saving 20 minutes of contributor time is minimal. Secondly, the editorial quality that Independent Reviews aims to deliver is fundamentally the assessment of named, credentialed humans who have conducted the testing themselves. AI-generated text undermines that premise even when the writing is technically accurate. Thirdly, the long-term credibility of the publication relies on the readers' trust that a byline corresponds to a real individual who has read the studies they are referencing; we do not wish to compromise that reputation for short-term gains.

Disclosure on individual pieces

When an AI tool is utilized in any of the three approved manners for a specific piece, we currently do not disclose the specific usage because the use is so generalized (all contributors use these tools as aids for research) that a piece-by-piece disclosure would not provide meaningful information. If we decide to alter this policy, for example, if we publish AI-generated draft content, even with extensive editing, we will announce the change here and mark the affected pieces individually.

This policy is dated

Last updated April 21, 2026. AI tools evolve rapidly; we plan to reassess this page at least quarterly and to note significant changes in our update log.