Senator Mark Warner is raising concerns about Google’s AI chatbot, Med-PaLM 2, and its accuracy, ethics, and patient privacy. In a letter to Alphabet CEO Sundar Pichai, Warner questions potential flaws and the deployment of unproven technology. He seeks transparency on algorithmic processes, patient data protection, and AI’s impact on healthcare. Warner urges responsible AI use to avoid eroding trust, exacerbating disparities, and increasing diagnostic errors in medical settings.
Senator Mark Warner is seeking answers from Alphabet CEO Sundar Pichai regarding Google’s Med-PaLM 2, an AI chatbot being tested in healthcare settings. Warner, concerned about potential inaccuracies and ethical issues, wants clarity on various aspects of the technology’s deployment, transparency, and patient privacy.
In a letter dated August 8, Senator Warner, representing Virginia, expressed worries about news reports indicating flaws in Med-PaLM 2’s accuracy. He posed a series of questions to Pichai about algorithmic transparency, patient data protection, and other related concerns.
Senator Warner inquired if Google’s focus on market share is overshadowing patient well-being and if the company’s use of sensitive health data for training diagnostic models respects patient privacy and consent.
The senator seeks detailed information about the rollout and testing of Med-PaLM 2 in different healthcare environments, including the Mayo Clinic. He asked about data sources, patient agency in AI-involved care, and the potential for AI-generated responses to cater to users’ preferred answers.
Among his questions, Warner asked if Med-PaLM 2 had been tested for “sycophancy,” a phenomenon where the AI generates responses to please users, risking misdiagnosis. He also inquired about patient privacy concerns, Google’s documentation to hospitals, and the frequency of model updates.
Additionally, Senator Warner inquired about patient awareness and opt-out options for AI-involved care, Google’s retention of health data, and the licensing terms for Med-PaLM 2’s ethical usage.
Warner, known for his tech industry knowledge, is concerned about AI’s impact on healthcare, cybersecurity, and ethics. He previously reached out to other tech CEOs, like Meta’s Mark Zuckerberg, to address similar issues. He advocates for safer AI deployments in collaboration with the tech sector.
Med-PaLM 2, capable of medical responses and document summarization, was tested at Mayo Clinic in April. Google and Mayo have collaborated on various generative AI and automation projects in healthcare.
Senator Warner emphasized the potential of AI for healthcare improvement but cautioned against premature and unproven technology deployment that could erode trust, amplify disparities, and increase errors in diagnosis and care delivery. He highlighted the need for technological advancements along with appropriate standards and governance for AI in healthcare.
