AI outputs are fictions, not hallucinations, experts argue

By PulseAugur Editorial · [1 sources] · 2026-08-13 07:54

The term "hallucination" is inaccurate when describing the output of large language models. Instead, these outputs should be considered "fictions," as every word generated by a trained model is essentially fabricated. While these fictions may incorporate elements of reality, making them appear more convincing, their fundamental nature remains artificial. AI

IMPACT Challenges the common understanding and terminology surrounding AI model outputs, suggesting a re-framing of how we discuss AI-generated content.

RANK_REASON Opinion piece arguing against the common terminology used to describe AI output.

Read on Mastodon — fosstodon.org →

opinion

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

AI outputs are fictions, not hallucinations, experts argue

COVERAGE [1]

Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] · 2026-08-13 07:54

They're not "hallucinations", they're fictions. Every word that comes out of a Trained # MOLE is fiction, just like every image and video. When it mixes in elem

They're not "hallucinations", they're fictions. Every word that comes out of a Trained # MOLE is fiction, just like every image and video. When it mixes in elements of the real world, as all fiction does, that might make it more convincing, but it doesn't change what it is. # All…

COVERAGE [1]

They're not "hallucinations", they're fictions. Every word that comes out of a Trained # MOLE is fiction, just like every image and video. When it mixes in elem

RELATED TOPICS