Back to the catalogue

Governance and risk

Bias Detector (Does your AI treat everyone the same?)

An AI application can treat some people worse than others without anyone wanting it. The zip code may be telling someone's origin without anyone ever asking. This tool walks with you through seven blocks and forty questions from a simplified version of the official Adigital and IAméricas bias analysis and mitigation checklist, and adapts the depth to what you really control. If you only use a third-party tool, it asks you for what is in your power and turns the rest into questions for your provider. If you develop the model, it asks you for evidence, metrics and thresholds. When you close each block, it gives you the hot discovery, so the conversation works for you even if you interrupt it in the middle. She is honest about her reach. The original checklist is designed for an advanced training program with expert support. Applied in a conversation and without verified evidence, it gives an indicative photo.

iamericas-detector-sesgos
Download skillv1.2

Free download, no sign-up

What problem does it solve?

The skill starts by asking you for a free description of the system. From there he deduces your profile and confirms it in a sentence, always in terms of what he is going to ask you and what he is not going to ask you, never labeling you. Then go through the blocks that apply. Group the questions, reuse what you already said and translate your answer into one of the official options, confirming it before writing it down. If you doubt, explain the concept with an example from your own case before asking again. Only mark that you don't know after trying to help you, not at the first hesitation. If you are already concerned about a specific bias and your profile supports it, it offers you to delve deeper into that axis so that the recommendations are concrete and not generic.

What does it produce?

A report in Word and PDF with two scores, the general one and the one within your scope of control.
  • The findings classified into five categories. Actionable gaps, pending to be discovered, partial, beyond your control and not applicable.
  • Between three and five concrete actions, prioritized by impact and effort, anchored in a real finding and not in generic advice.
  • The list of questions for your supplier, written to copy and send, with the specific document that you have to request in each case.

When to use it

  • Your AI involves people. Personnel selection, credit, pricing, customer service or content moderation.
  • Someone has complained about unequal treatment of you.
  • You are going to expand the use of a system that already works to more people or to another group.
  • You want to prepare serious questions for your supplier and not have them answer you with generalities.
  • Your client or your sector begins to ask you for guarantees of equity.

What you need first

  • What the system does, who it affects, and what it decides or recommends.
  • If you bought it ready, you integrated it or trained it yourself.
  • What data do you use to decide.
  • If there is a person reviewing the results before they take effect.
  • If you are already concerned about a specific bias, due to gender, age, origin, disability or socioeconomic level.
  • You do not need to bring real data on the affected people. Categories and examples are enough.

Where it works and how to install it

Pick your platform and follow the steps. Where the tab is labelled “Native” the file installs as-is; ChatGPT and Gemini do not install it yet, so there you paste the content of SKILL.md into the chat.

Method A — Upload the file (easiest)

  1. 1Download the file with the button above. No need to unpack it.
  2. 2Open Claude — the desktop app on Windows or Mac, or claude.ai — and go to Customize → Skills.
  3. 3Click “Add” and upload the file as it is. If the form rejects the extension, rename .skill to .zip and upload again: same file.
  4. 4The skill shows up in your capabilities list, ready to switch on.

It is installed. Turn it on from the capabilities panel, or type @ followed by its name inside any conversation.

Version history

  • v1.2

    21 September 2026 · 795 kB

    Changes how the conversation starts. If the first message brings no context, the skill explains how the process works before asking anything: the phases, that it can be paused and resumed, and what you get at the end; and it notes that the input can be typed or attached as a file. If the context is already there, it confirms in two lines and gets going without that explanation. The description is rewritten to state that this is a guided, step-by-step process.

  • v1.1

    14 September 2026 · 791 kB

    Neutral Spanish throughout. It works out which Python interpreter the system has and warns up front when the dependency the Word report generator needs is missing.

  • v1.0

    1 August 2026 · 786 kB

    First version.

Other skills on the same topic