The most dangerous lie AI keeps repeating | Karen Colbert | TEDxBellarmineU

Quick Overview

The most dangerous lie artificial intelligence keeps repeating is the assumption that it is neutral, which masks the inherent biases and historical exclusions present in the data it is trained on, ultimately diminishing the authority of marginalized communities like Indigenous peoples.

Key Points: The speaker, Karen Colbert, identified the dangerous lie that Artificial Intelligence (AI) is neutral, arguing that it repeats historical biases found in its training data. Colbert launched her dream course on Artificial Intelligence analysis and history in the summer of 2026. Tribal colleges, such as Kiiwataw Ojiibwe Community College, are accredited institutions explicitly focused on the preservation, protection, and revitalization of Indigenous language, land, and identity. The core issue is that AI models trained on existing data reflect and often accelerate existing societal inequities, rather than being objective tools. Colbert realized the moment that AI's confidence, often sounding fluent and polished, is not the same as truth, as it lacks genuine understanding of context. The speaker urges the audience to recognize their shared responsibility to reject the notion that AI is inherently neutral and to actively ensure Indigenous voices are not excluded from digital spaces. The realization struck Colbert when an AI output she received was not a representation of AI failure, but a reflection of the biased data it was trained on.

Context: Karen Colbert, a data scientist and General Education Department Chair/Lead Math Faculty at Kiiwataw Ojiibwe Community College, delivers a TEDx talk from the TEDxBellarmineU event. Her presentation focuses on the inherent biases within Artificial Intelligence (AI) systems, particularly how training data, often derived from historically dominant narratives, leads AI to perpetuate exclusion and misrepresentation of marginalized groups, specifically focusing on Indigenous communities.

Detailed Analysis

Karen Colbert argues that the most dangerous assumption surrounding Artificial Intelligence (AI) is its perceived neutrality. She recounts an early experience in 2026 when she launched a course on AI analysis and history, where she initially trusted AI outputs, even when they seemed flawed, mistaking their fluent presentation for accuracy. This changed when she recognized that AI systems, trained on existing data, inherently absorb and repeat societal biases and historical exclusions, such as the erasure of Indigenous culture, language, and identity. Colbert cites the work of tribal colleges, like Kiiwataw Ojiibwe Community College, whose mission is explicitly dedicated to the preservation and revitalization of Indigenous heritage, contrasting this lived reality with the 'glowing blue figures' and 'glossy, fictional images' often generated by AI. She stresses that AI's confidence in its output is not equivalent to truth or understanding, and that the responsibility lies with humans to actively partner with AI to mitigate bias and ensure that communities historically excluded from digital spaces are included in the data and the resulting narratives.