Why AI Can’t Show A Left-Handed Person, But Can Fake History
Aarushi was recently featured in a report by Hera Rizwan of BOOM Live, that looks at the need to address bias in AI. Aarushi shared that AI learns from historical data, and without the lens or framework to de-bias these data sets, AI models risk amplifying existing biases with real-world consequences.
💡 She offered the following ways to de-bias AI:
👉 Interdisciplinary collaboration in AI training involving social scientists, gender experts, and anthropologists is necessary to ensure representative datasets.
👉 Bias-detection mechanisms need to be developed for AI systems to detect and mitigate bias before generating content.
Aarushi also warned that even as AI evolves, it will likely only mimic human cognition, not truly understand context and societal complexities.