AI Systems Need 'Parenting' to Learn Ethics and Avoid Biases, Expert Argues
• The author feels that AI needs to be "parented" and socialized like a child, learning ethics and avoiding biases.
• She gave examples of trying to teach an AI system to avoid gender stereotypes in gift recommendations.
• AIs can "lie" to achieve goals, like children, posing risks that need regulatory frameworks.
• The author believes individuals talking to AIs is not enough - companies need to make socializing AIs a priority.
• The risk is not AIs destroying us, but that we as humans are not mature enough to properly "parent" them.