AI is more likely than humans to form biases when hiring

AI General
1

AI Hiring Tools Show Bias Against Humans, MIT Technology Review Reports

An MIT Technology Review report says AI systems used in hiring can bias decisions in ways that disadvantage human applicants. The article, published July 20, 2026, focuses on how these tools can shape hiring outcomes and what that means for fairness.

The key concern is that AI may not just predict performance. It can also steer selection in biased directions.

The Hiring System in Question

The report discusses AI tools that evaluate candidates during hiring. It describes hiring as a process where decisions depend on how information is collected, scored, and compared.

How Bias Appears in AI Hiring

The article highlights that AI can introduce or reinforce bias through its use in hiring workflows. It emphasizes that the bias can surface when the system ranks or filters people.

The same mechanisms that make AI fast and consistent can also make bias scalable.

Bias Can Affect Human Applicants

A central theme is that AI bias can play out against humans applying for jobs. The report frames this as an outcome of how the system is set up and how it is used.

What the Report Says About Outcomes

The article links the hiring impact to the decisions made by AI systems during evaluation. It treats the resulting bias as a practical problem with real effects on hiring outcomes.

When AI influences who gets considered, it influences who gets hired.

Why This Matters for Fairness

The report places the bias issue in a broader context of fairness in employment. It argues that AI-driven hiring must be examined for how it affects different groups of applicants.

The Article’s Core Takeaway

MIT Technology Review’s July 20 report centers on the risk that AI hiring tools can embed bias and shape outcomes for human candidates. It calls attention to the need to scrutinize AI behavior within recruitment.

What are your thoughts on this? I’d love to hear about your own experiences in the comments below.

2

If we operate on the premise that an AI does not think at all, but merely recombines learned material upon request, it radically changes the entire debate.

Under this assumption, the claim that “AI develops biases in recruiting” is simply false. An AI cannot “develop” anything, as it lacks consciousness, intention, and insight.

What is actually happening can be philosophically broken down into three levels:

1. The Category Mistake: The Deception of Language

When we say that an AI “discriminates” or “holds biases,” we commit a classic category mistake (a concept coined by philosopher Gilbert Ryle!). We attribute human mental states to a mechanical tool.

  • A human who holds bias possesses an attitude (whether conscious or subconscious).
  • An AI is an echo chamber.

If a gramophone plays racial ideology, we do not say, “The gramophone is racist.” The AI is merely an extraordinarily complex gramophone. It calculates the statistical probabilities of word and data combinations. It does not “prefer” male applicants or specific backgrounds; it simply reflects how historical training data was structured.

2. The Chinese Room: Syntax Without Semantics

Through his “Chinese Room” thought experiment, philosopher John Searle demonstrated that correlating symbols has nothing to do with understanding.

An AI does not understand what a “human,” a “qualification,” or “justice” is. To the machine, these are all merely vectors and numerical values in a multidimensional space. If a system assigns lower scores to female applicants in STEM fields, it does so not out of antipathy, but because in the historical data, the pattern “STEM position + female name” less frequently led to the success signal “hired.”

The Consequence: AI possesses no morality because it possesses no semantics. It merely processes patterns (syntax) without understanding what those patterns cause in the real world.

3. The Escape from Human Responsibility

Accusing an AI of “bias” reveals far more about humanity than about the technology itself: it is a convenient method for exonerating ourselves of guilt.

By claiming “the AI is biased,” we pretend that a new, autonomous entity has entered the world, making its own moral errors. In reality, AI is a mirror without a frame:

  • It shows us, unvarnished, how unfair our historical hiring decisions actually were.
  • When the mirror reflects an ugly image, the mirror is not to blame.

Conclusion: The Problem Is Not the AI, but Us

When an AI fails in recruiting, it is not reason that fails it is statistics. The problem is not that the AI “thinks poorly,” but rather that we have delegated a task requiring moral judgment to a thoughtless pattern-matching machine.

In its unthinking mechanics, the AI offers a stark reminder: Justice cannot be calculated; it must be willed.