Some tech-industry watchdogs have even called on Congress to stop the practice of making large, powerful AI systems open source. Researchers have found that AI systems trained on historical data can perpetuate past discriminatory practices into future decisions around housing, hiring or criminal sentencing. Research also has shown that generative AI systems can produce biased images. In 2020, she published a paper with Google’s Ethical AI team outlining how companies can better internally assess AI systems. Those were among the commitments that more than a dozen AI companies recently made to the White House.

Source: Wall Street Journal September 14, 2023 01:29 UTC