📱 AI Is Still Underhyped: Ex-Google CEO Eric Schmidt

In a candid TED interview, former Google CEO Eric Schmidt breaks down what's holding AI back—and what risks we're underestimating. Below are key takeaways without the panic but with a sharp focus.

💬 AlphaGo's victory sparked a revolution. In 2016, DeepMind's model made a move no human had conceived in 2,500 years. That was the wake-up call: AI can invent, not just imitate.

💬 AI is still underhyped. ChatGPT is just the shop window. Recent breakthroughs in reinforcement learning, planning, and strategy enable AI to independently perform complex, multi-step tasks.

💬 AI can recombine what it knows but struggles with true discovery. Unlike human scientists, it can't yet make paradigm-shifting, cross-domain insights. The next challenge is building systems that can invent their own goals and frameworks.

💬 Three major bottlenecks are holding AI back:

— Energy: The U.S. alone may soon need 90 gigawatts of power (about 90 nuclear plants) just to support AI data centers.
— Data: We're close to exhausting useful training datasets. Synthetic data will be necessary.
— Compute: Switching from deep learning to full-scale reinforcement learning may require up to 1,000× more computational power.

💬Geopolitical risks. AI is dual-use: civilian and military. The race is on between the U.S. and China. The first to reach superintelligence could gain uncontested technological dominance, destabilizing global security. Schmidt compares it to nuclear deterrence but with even less margin for error.

"If you're not using AI, you're going to fall behind. You should be using it every day. This is not episodic. This is a marathon," says Schmidt.

📱 Watch the full TED talk here.

More on the topic:

🎤 Why future generations might pity us: TED talk by Sam Altman

🎤 What if AI isn't making our lives better—but worse?

#news #interview @hiaimediaen