Thinking Space

A classic example comes from the time of nuclear race between USA and USSR.By 1983, the Soviets had built a brand new early warning system called Oko. Ready to detect the very first launch of an American missile attack. A state-of-the-art system, it did away with the possibility of human error. On 26 September 1983, Stanislav Petrov was in charge of Oko. He was a lieutenant colonel in the Soviet Strategic Air Defence Forces. At half past midnight, a warning light came on and a siren started screaming. Oko showed a missile being launched from the USA. Everyone in the control room froze. Then a second warning light and siren showed another missile. Then a third missile. Then a fourth missile. Then a fifth. This was an all-out nuclear attack on the USSR. Petrov had clear orders. Pick up the phone and order a nuclear response. If he waited too long, the American missiles would hit and destroy all the Soviet missile bases. Their country would be utterly destroyed. Millions upon millions dead, and those left alive at the mercy of the Americans. His orders were clear. But Stanislav Petrov didn’t follow the orders. He sat and thought for a minute. And he did the unthinkable. He questioned the foolproof computer system. And he allowed human reasoning to override the computer. He thought, Why only five missiles if it’s an all-out attack? And he decided the computer was wrong. He refused to pick up the phone and report an attack. As the seconds crawled by, his uniform became drenched in sweat. Everyone in the control room watched him and held their breath. Until the truth was confirmed. Ground radar and geostationary satellites reported there were no American missile launches. The supposedly foolproof system had misread a one-in-a-million alignment of the sun’s rays on high-altitude clouds.

Since then computers have become much more faster, more accurate and smarter. The whole movement of machine learning and deep learning is taking the computers to next level but at the same time the world is becoming much more complex and hence they will still need the experience and intuition of humans.