'AI testing today is more politics than science': Rumman Chowdhury
Rumman Chowdhury, CEO of Humane Intelligence and former US Science Envoy for AI, says today's AI testing is more about politics than real science.
She argues that current methods are based on personal opinions instead of solid evidence, which she says can leave AI systems fragile and misaligned with people's needs.
Chowdhury directed Twitter/X's ML Ethics, Transparency, and Accountability team, which was dissolved after Elon Musk's acquisition.
Institutionalizing AI safety, regulation, and testing
Chowdhury is focused on turning big ideas about AI ethics into real guidelines for institutionalizing safety, regulation, testing and risk frameworks.
Her goal: make sure the way we use AI is transparent, accountable, and actually serves communities—not just tech companies.
AI can confidently provide wrong answers
Chowdhury also warns against putting too much faith in AI.
She points out that these models can confidently give wrong answers or even sway people if prompted assertively.
Just because something is data-driven doesn't mean it's neutral—AI still reflects human biases and isn't automatically trustworthy or beneficial.