Bias & Discrimination: The AI Danger Explained

What is the danger?

AI learns from historical data, and history is full of discrimination. When AI decides who gets hired, who gets a loan, who gets flagged by police or how sick a patient is, it can repeat and amplify that discrimination — while looking objective.

How it happens

Training data that under-represents some groups.
Systems that learn to use proxies like postcode or name for race or gender.
Decisions that cannot be explained, so they are hard to appeal.

Real cases

Amazon scrapped an experimental AI recruiting tool after finding it downgraded CVs that included the word “women’s” (reported 2018).
Source: Reuters ↗
Robert Williams was wrongfully arrested in Detroit in 2020 after a facial-recognition system falsely matched him to a shoplifting suspect.
Source: ACLU ↗
A 2016 investigation found a widely used criminal risk-score algorithm (COMPAS) falsely flagged Black defendants as future criminals at nearly twice the rate of white defendants.
Source: ProPublica ↗

Warning signs

An automated rejection with no human you can contact.
Decisions described as “the system decided”.

What you can do

Ask whether AI was used in a decision about you, and request a human review.
Support laws that require bias audits for high-stakes AI.

Frequently asked questions

What is the danger of bias in AI?

How does it happen?

Training data that under-represents some groups. Systems that learn to use proxies like postcode or name for race or gender. Decisions that cannot be explained, so they are hard to appeal.

How can I protect myself?

Ask whether AI was used in a decision about you, and request a human review. Support laws that require bias audits for high-stakes AI.

Bias & Discrimination