Are Chatbots Smarter Than Doctors? The Real Truth Behind AI in Healthcare

In a 2024 year-end article, The Washington Post encouraged readers to be thankful for AI in medicine, praising the diagnostic accuracy of AI chatbots over traditional doctors. This optimism reflects a growing trend where Artificial Intelligence, particularly in healthcare, is perceived as a revolutionary force. With AI chatbots reportedly diagnosing with 90% accuracy—higher than the 74% of doctors—the future of AI in medicine appears promising. But is the hype justified?

Nandan Nilekani: Higher Standards for AI, Rightly So

Infosys co-founder Nandan Nilekani has pointed out a key issue with public perception: while human errors are routine (as with car accidents), even minor AI failures invite heavy scrutiny. This, he argues, stems from the fact that AI, unlike humans, doesn’t possess understanding or accountability—meaning it cannot be held responsible in the same way. This lack of agency makes blind trust in AI technologies dangerous, especially in life-or-death areas like healthcare.

Also Read: Everything You Need to Know About the Blue Aadhaar Card for Children

Can AI Replace Doctors? Bill Gates Thinks So, Critics Disagree

Some, like Bill Gates, believe AI could eventually replace medical professionals. However, this would require AI systems to exhibit human-like reasoning or consciousness—something far beyond current capabilities. Even with high diagnostic accuracy, AI lacks contextual understanding, empathy, and dynamic decision-making skills necessary for complex healthcare environments.

The Limitations of LLMs: What the Experts Are Saying

Today’s AI revolution is largely driven by Large Language Models (LLMs) like ChatGPT, Gemini, and LLaMA. But according to AI pioneer Yann LeCun, these models may become obsolete soon. He urges developers to focus on next-gen systems that overcome current limitations, such as hallucinations and biases.

Prominent AI critics like Gary Marcus and cognitive scientists Arvind Narayanan and Sayash Kapoor argue that these issues are inherent to LLMs and not easily fixable. They highlight that LLMs merely recognize patterns in training data—they don’t understand meaning or truth. This makes them prone to fabricating information, no matter how large or well-curated the dataset is.

The Hidden Human Cost of Training AI

Another issue the hype often masks is the human labor involved in AI training. Labeling vast amounts of toxic or sensitive data is often outsourced to developing nations where labor is cheaper and regulations weaker. This process can be traumatic and exploitative for workers. The AI boom, therefore, isn’t just a technological story—it’s a socio-economic one too.

AI Hype and the Real Risks We Must Confront

According to Marcus’s book Taming Silicon Valley, there are three main strategies tech companies use to fuel AI hype:

1. Unfounded claims about reaching Artificial General Intelligence (AGI)
2. Invoking China as a competitive threat
3. Promoting doomsday or utopian narratives about AI’s impact

These strategies drive up stock prices and secure government funding while diverting attention from real and current threats like misinformation, cybercrime, and the exploitation of labor.

Responsible Innovation, Not Sensationalism

AI undoubtedly holds transformative potential in fields like medicine. But as critics argue, its current capabilities are often exaggerated. If AI is meant to improve efficiency in isolated, repetitive tasks, its development must be balanced against public investment needs in areas like life sciences. If the goal is to replicate human intelligence, then governments and corporations must provide clear, realistic roadmaps—and not just dreams fueled by marketing hype.