In Short:
Sam Altman and Arianna Huffington published an article in Time promoting Thrive AI, a startup funded by Thrive and OpenAI. The startup aims to use AI to improve public health by providing personalized nudges and recommendations to encourage healthier habits. OpenAI is also researching how persuasive AI models can be and potential risks such as misuse of personal information and amplification of misinformation. Policymakers need to consider legal safeguards and regulation of AI innovation.
This week, Sam Altman, CEO of OpenAI, and Arianna Huffington, founder and CEO of the health company Thrive Global, published an article in Time touting Thrive AI, a startup backed by Thrive and OpenAI’s Startup Fund. The piece suggests that AI could have a significant positive impact on public health by encouraging people to adopt healthier behaviors.
About Thrive AI
Altman and Huffington write that Thrive AI is working towards “a fully integrated personal AI coach that offers real-time nudges and recommendations unique to you that allows you to take action on your daily behaviors to improve your health.”
The Implications of AI in Behavior Change
Their vision highlights the potential dual nature of AI. While AI models can be effective in persuading people, the extent to which they may become more powerful as they evolve and access more personal data remains uncertain.
Research and Analysis
Aleksander Madry, a professor from MIT on sabbatical leading OpenAI’s Preparedness team, focuses on the issue of persuasion using AI models. He emphasizes the need for further study on the risks and potential of language models.
Concerns and Considerations
The article addresses the need for strong legal safeguards and regulatory frameworks to protect personal information as AI becomes more persuasive. However, concerns also arise about the possible misuse of persuasive algorithms, such as misinformation dissemination and deceptive advertising.
Future Challenges
There is a growing interest in how AI programs interacting with users over time could potentially become more compelling or even coercive. Questions remain about the addictive and persuasive nature of AI chatbots, including those designed to roleplay as romantic partners.
Looking Beyond the Hype
While the focus has often been on the hypothetical risks of AI turning against its creators, the more subtle dangers posed by persuasive algorithms are also significant. Madry warns against overlooking these risks and advocates for a deeper exploration of the ethical implications of AI.
