Archive · 14 Jun 2025
AI's Consequential Decisions: Concerns and Future Policies
Imagine AI as a tool that's supposed to make smart choices for us, like deciding who gets a loan or how to treat a patient.
But Sayash Kapoor, who wrote a book about AI, warns that some AI tools aren't as reliable as they seem, especially when making big decisions.
He explains that AI can struggle because it doesn't always have enough information, like knowing how someone truly feels.
He also shares how the government is concerned about the influence of AI on jobs and digital images that have been altered.
He encourages us to train people to use the AI and manage AI responsibly, to prevent problems with altered pictures, and use AI to help improve our lives.
Sayash Kapoor, co-author of "AI Snake Oil," discusses concerns about predictive AI in crucial decisions.
Kapoor highlights potential risks of AI in loan approvals, law enforcement, healthcare prioritization, and education.
He emphasizes limitations of predictive AI, citing its low accuracy in real-world applications and insufficient data.
Kapoor observes a shift in government focus from existential AI risk to impacts on jobs, deepfakes, and information environments.
He suggests policy priorities for India, including workforce training and addressing the proliferation of deepfakes.
- Who
- Sayash Kapoor, author of "AI Snake Oil", and Venkatesh Kannaiah of indianexpress.com.
- What
- An interview with Sayash Kapoor about the impacts and concerns of AI.
- When
- The interview took place recently.
- Where
- The interview was conducted by indianexpress.com.
- Why
- To discuss the current state of AI, its potential risks, and policy recommendations.
- Expert
- Sayash Kapoor
- Book Title
- "AI Snake Oil: What Artificial Intelligence Can Do, What It Can’t, and How to Tell the Difference"
- Primary Focus
- Societal impact of AI, especially predictive AI
- Key Concern
- AI's reliability in consequential decision-making
- Government Focus Shift
- From existential risk to impact on jobs, deepfakes, and information environment
- Policy Recommendation
- Workforce training and addressing deepfakes