It’s the Second AI Singularity that Should Alarm Us

We had no reason to be alarmed when we built machines that are stronger than our biological self. Why then should we be overly alarmed by machines that can out-think us? Machine already grossly out-compute us, out-search us, out-remember us, let them also reason deeper and better, no survival alarm there. The specter to pose a threat is the one where machines are so smart that they can develop means to dupe us, to cause us to surrender the keys, to act against our survival interests. Alice may be much smarter than Bob, but not be smart enough to lure Bob to jump off the bridge, thinking he walks in the park. As long as Bob has his survival wits about him, he would only benefit from a smarter companion. But when Alice gets so smart that she can implant thoughts in Bob’s mind, then the alarm bells should ring. This is the second AI singularity.

Fortunately, we have good defenses against this threat. It’s called randomness. As smart as machines can become, they can never defeat randomness. We should surround ourselves with this protective shield. How to do it is a topic for another blog. The general idea is here