The current AI boom will amplify social issues.

2023-05-23

If we don't take action now, the current AI boom will amplify social issues, according to AI ethicists. AI systems trained on internet data can potentially reinforce negative biases and cause social harm. Salesforce's Kathy Baxter explains how to address this issue. AI developers must act quickly to develop and deploy systems that address algorithmic biases, says Kathy Baxter, Chief Architect of AI Ethics at Salesforce. In an interview with ZDNET, Baxter emphasizes the need for diversity and representation in datasets and user research to ensure fair and just AI systems. She also highlights the importance of transparency, interpretability, and accountability in AI systems while protecting individual privacy. Baxter emphasizes the need for cross-department collaboration, similar to the model used by the National Institute of Standards and Technology (NIST), to develop powerful and secure AI systems that benefit everyone. One of the fundamental issues in AI ethics is ensuring that the development and deployment of AI systems do not reinforce existing social biases or create new ones. To achieve this goal, Baxter emphasizes the importance of asking who benefits and who pays for AI technology. Considering the datasets being used and ensuring they represent the voices of everyone is crucial. Inclusivity in the development process and identifying potential harms through user research are also essential. "This is one of the fundamental questions we must discuss," says Baxter. "Especially women of color who have been asking this question and doing research in this field for years. I'm glad to see many people discussing this issue, especially with generative AI. But fundamentally, what we need to do is ask who benefits, who pays for this technology. Whose voices are included?" Social biases can be injected into AI systems through the datasets used to train them. Non-representative datasets that contain biases, such as image datasets primarily focused on one race or lacking cultural diversity, can lead to biased AI systems. Additionally, the unequal application of AI systems in society can perpetuate existing stereotypes. To make AI systems transparent and understandable to the general public, prioritizing interpretability during the development process is crucial. Technologies like "explainable AI" can help AI systems demonstrate their workings and make the decision-making process more understandable. User research is vital in ensuring clear explanations and enabling users to identify uncertainties in AI-generated content. Protecting personal privacy and ensuring responsible use of AI require transparency and permission. Salesforce follows the guidelines of responsible AI generation, which includes respecting the sources of data and using customer data only with their consent. Allowing users to choose to join, opt-out, or control their data usage is crucial for privacy. "We only use customer data with their consent," says Baxter. "When you use someone's data, be transparent, allow them to opt-in, and allow them to opt-out when they say they don't want their data included. That's really important." As the competition for AI innovation intensifies, maintaining human control and autonomy over increasingly autonomous AI systems is more important than ever. Empowering users to make informed decisions about the use of AI-generated content and keeping humans informed helps maintain control. Ensuring the security, reliability, and availability of AI systems is crucial, and industry-wide collaboration is essential to achieve this goal. Baxter praises the AI risk management framework created by NIST, which involves over 240 experts from various industries. This collaborative approach provides a common language and framework for identifying risks and sharing solutions. Failure to address these AI ethics issues can have serious consequences, such as facial recognition errors or the generation of harmful images leading to wrongful arrests. Investing in safeguards and addressing current, not just future, potential harms can help mitigate these issues and ensure responsible development and use of AI systems. While the future of AI and the possibility of general AI are intriguing topics, Baxter emphasizes the importance of focusing on the present. Ensuring responsible use of AI and addressing social biases today will prepare society for future advancements in AI. By investing in AI ethics practices and cross-industry collaboration, we can help create a safer and more inclusive future for AI technology. "I think the timeline is really important," says Baxter. "We really have to invest now, create that muscle memory, create those resources, create regulations that allow us to move forward but move forward safely."

RECENT AI NEWS

RECENT AI TOOLS