Reflection on Eric Schmidt’s TED Interview

My Personal Views and Reflections

1. On the Speed and Implementation of the AI Revolution

Although Eric Schmidt is very optimistic that the AI revolution will bring about tremendous changes within five years, possibly even triggering global geopolitical risks, I tend to believe that the transformation brought by AI is a “decade-level” gradual process. As Andrej Karpathy mentioned in the 2025 Y Combinator AI Startup School with the example of autonomous driving, there is often a long “engineering gap” between technological breakthroughs and large-scale implementation. The complexity of the real world, engineering challenges, policy barriers, and social adaptation all mean that AI adoption will not happen overnight. Scientists’ way of thinking focuses more on practical feasibility and system robustness, rather than simply extrapolating trends. The revolutionary potential is undeniable, but true popularization often takes much longer.

2. Resources Are Never Enough

I strongly agree with Eric’s view that “resources are never enough.” Whether it’s computing power, data, or energy, whenever there are more resources, scientists and engineers always find new problems and methods to use them up. When I was doing DFT molecular simulations during my PhD, limited computing resources meant I could only do a small number of calculations. But even now, with abundant resources, more complex needs quickly exhaust them. Innovation and demand always make resources “never enough.”

3. The Black-Box Nature and Interpretability of AI

The “black-box” nature of AI makes me uneasy. While some people think “as long as it works, it’s fine,” I prefer to understand the underlying principles—only by truly understanding how AI works can I use it with confidence. Neural networks are far more powerful than our understanding of their principles. Science pursues interpretability and controllability, which is also the direction AI must break through in the future.

4. AI Barriers and Industry Landscape

Although companies like OpenAI have established certain technical barriers through closed-source approaches, in reality, technology diffusion and innovation happen very quickly. Teams like Google, Anthropic, and DeepSeek, with strong R&D capabilities and resources, can quickly catch up or even surpass. Barriers in the AI field mainly lie in computing power, data, talent, algorithmic innovation, ecosystem, and capital, but none of these are insurmountable. The industry landscape is still full of uncertainties and far from a “winner-takes-all” situation.

5. AI’s Interaction with the Physical World

I believe a core barrier to AI capability improvement is the lack of real interaction and experience with the physical world. Most AI today operates in virtual environments or simulations. Only when AI can deeply interact with the physical world, explore autonomously, and learn through trial and error, can true breakthroughs be achieved. AI is like a brain without a body; only when combined with sensors, robots, etc., can it unleash greater potential.

6. Distinguishing Real from Hype in AI Applications

Faced with the endless stream of new AI products and “hype” every day, I think we should focus on those that truly solve real problems, are widely adopted, continuously evolve, and deeply integrate with industries. Maintain critical thinking, experience personally, track over the long term, and avoid being misled by short-term hype. Letting AI “land” and making AI “useful” are the directions truly worth investing in.

7. My Ideal Direction for AI

I am most interested in AI applications in health, longevity, and life sciences. I hope to use AI to reveal the essence of life, disease mechanisms, and aging processes, pushing the limits of human health and longevity. Human understanding of our own biology is still very limited. AI is expected to become a powerful tool for breakthroughs in biology. In the next decade, AI + biology is likely to bring unprecedented scientific advances.

Summary:
The AI revolution is full of hope, but also full of challenges. We must be wary of short-term hype and extreme risks, while maintaining long-term thinking and rational reflection. Continuous learning, embracing change, focusing on implementation, and pursuing the essence are my core attitudes in facing the AI wave.