AI is great for rich and powerful people and for tech giants trying to boost profits. Otherwise, artificial intelligence and the automation it enables can be harmful, nonprofit Mozilla concluded in a report published Monday.
"In real life, over and over, the harms of AI disproportionately affect people who are not advantaged by global systems of power," Mozilla researchers conclude in the 2022 Internet Health Report. "Amid the global rush to automate, we see grave dangers of discrimination and surveillance. We see an absence of transparency and accountability, and an overreliance on automation for decisions of huge consequence."
AI, systems trained with vast swaths of complex real-world data, is revolutionizing computing tasks that were previously difficult or impossible. That includes recognizing speech, spotting financial fraud, piloting self-driving cars, identifying birds by their songs, looking for. As AI spreads into every corner of tech, though, experts are raising concerns about its problems, too.
Mozilla, the nonprofit that builds the Firefox web browser and advocates for privacy on the web, is among those critics. The AI problems Mozilla spotlighted include:
- Machine learning models often reproduce racist and sexist stereotypes because of bias in the data they draw from, including internet forums and photo archives.
- Big companies aren't transparent about how they use our personal data in the algorithms that recommend social media posts, products to purchase, and so on.
- Recommendation systems can be manipulated to show propaganda or other harmful content. In a Mozilla study of YouTube, algorithmic recommendations were responsible for showing people 71% of the videos they said they regretted watching.
Companies like Google and Facebook have major programs for dealing with issues like AI bias. For example, Google is trying to, to ensure systems are better at representing people with darker skin.
But Mozilla doesn't like the fact that Big Tech funds a lot of academic research and that relatively few papers -- especially among those most widely cited -- focus on AI's social problems or risks.
Among Mozilla's suggestions are new laws. "Regulation can help set guardrails for innovation that diminish harm and enforce data privacy, user rights, and more," Mozilla said. Also on Monday, Mozilla released a five-part podcast on its concerns about AI.