Life 3.0

Points clés

• Life 3.0 means a form of life able to redesign both its software and its hardware
• Intelligence is about achieving goals, and it does not require a biological brain
• The real AI risk is competence with misaligned goals, not machines turning evil
• Explosive growth is possible because an AI could improve its own design
• The future is not fixed; what we want from AI is a choice we should make now

À l'intérieur de ce bit

1. Life that designs itself

   Imagine a company that quietly builds an AI able to write better code than its own engineers. Within months it dominates media, markets, and politics, not through violence but through sheer competence.

2. Intelligence is substrate independent

   Tegmark defines intelligence plainly as the ability to accomplish complex goals. Under that definition, there is nothing magical about carbon.

3. Goals matter more than malice

   Popular films picture killer robots with red eyes. Tegmark thinks that picture is a distraction.

4. Many futures are on the table

   Rather than predict, Tegmark maps possibilities. In one scenario, a benevolent AI runs the world so well that humans live in comfort but lose their sense of purpose.

5. Consciousness and what matters

   If machines become far more capable than us, does anything we care about survive? Tegmark turns to consciousness, which he describes as subjective experience, the fact that it feels like something to be you.

6. Passez à l'action

   This week, notice where AI already touches your life, from the route your maps app picks to the loans a bank approves, and ask what goal each system is really optimising. Read one plain-language piece on AI safety so you can discuss it with more than headlines.