Defense Department’s Efforts To Combat AI Bias Don’t Go Far Enough
This story originally was published by Real Clear Wire
By Delaney Duff
Real Clear Wire
Google’s difficulty in mitigating bias from their artificial intelligence systems – even after explicitly going to great lengths to minimize bias – spells trouble for the Department of Defense. Bias can cause AI tools to irrevocably malfunction and derail AI development. Google recently paused Gemini, their largest and most capable AI model, from generating images after it created historically inaccurate and offensive images of people. Google explained that their attempts to design a less biased, more inclusive image generation tool caused the model to malfunction instead.
This is especially concerning as the Defense Department plans to leverage AI at scale,
