Tier 3: Recommendations to Replace

Terms in this list should be considered for replacement. Terms included in this list have one or all of the following:

  • Review by the Inclusive Naming Initiative, with particular attention paid to finding consensus among member companies and participants’ companies for replacements
  • Research conducted by the Inclusive Naming Initiative on the etymology of the word and non-tech cultural connotations, per the Language Evaluation Framework
  • A consensus-based replacement term or terms proposed by the Language Workstream and sent for approval and review to the larger Initiative

Terms

Word List: Tier 3

Term: hallucinate

Definition

An artificial intelligence (AI) application generates information that has no basis in fact.

  • None

Usage Context

Recommendation

Recommended to replace when possible.

  • inaccurate information (noun)
  • *create* or *generate* inaccurate information (verb)

Unsuitable Replacements

  • None

Rationale

Hallucinations – perceptions that are not based in reality – are often associated with mental illness or drug use. Using the term in a technology context, in either its noun or verb form, can be seen as insensitive to people who experience those conditions.

Term Status

N/A

Supporting Content

N/A

Word List: Tier 3

Term: man-in-the-middle

Definition

N/A

  • None

Usage Context

Recommendation

Consider replacement

  • Adversary-in-the-middle attack
  • Interceptor attack
  • Intermediary attack

Unsuitable Replacements

  • None

Rationale

When an attacker secretly intercepts and relays communications between two systems or people who believe that they are communicating directly with each other. Gendered language that can perpetuate bias and stereotypes. Implies that women do not have the skills to perpetrate this type of hacking.

Term Status

N/A

Supporting Content

N/A

Word List: Tier 3

Term: Segregate

Definition

None

  • All Derivatives
  • Segregation

Usage Context

Recommendation

Replace

  • Segment/segmentation
  • Separate/separation

Unsuitable Replacements

  • Discriminate/discrimination

Rationale

The word segregation carries strong context in regard to civil rights movements in the US and South Africa, segregation in the US South, and racist history. Though the word appears etymologically neutral at first, Etymonline notes that the term has strong moral contexts prior to US segregation, and after US segregation is almost exclusively associated with the segregation of Black people from White people, an extremely racist context.

This meets one of the INI’s evaluation framework’s first-order concerns : the term is loaded, problematic, or politically charged outside of technology contexts, even if the language is itself etymologically neutral. As such, we recommend replacing it to remove the distracting, racist, and negative connotations of the word.

While the word is in use in security contexts and in GDPR and data protection contexts, it does not appear to be codified into any laws, policies, or other difficult to change or heavily embedded frameworks. Moreover, the replacement terms recommended—“segmentation” and “separation”—are both equally descriptive and in common use in technology, so we recommend replacing as you see the term.

We acknowledge that switching from “segregation” to “segmentation” or “separation” loses a small amount of nuance: specifically, “segregation” implies “separation” based on a policy or human-defined ruleset. If this is an issue in the context in which you use the word, we recommend using descriptive words along with the replacement, such as “policy-based segmentation.”

Term Status

N/A

Supporting Content

N/A

© 2023 The Inclusive Naming Initiative authors and collaborating organizations