menu

Ben Leggiero's Blog

Ben's opinions on Danger.

tl;dr AI is not a threat.

So many people are scared of artificial intelligence (AI). This fear comes from science fiction, and is as reasonable as fearing Klingons. The possibilities of AI are limited in scope and power. For instance, Google's most advanced AI can do things such as identify objects (cats, cars, etc.).[1]

Now, we have created a computer that can simulate a human brain![2] It took 82,944 processors and 1,000,000,000,000,000 bytes of memory 40 minutes to simulate a single second of activity in the brain. That's just the brain, mind you, without any senses, glands, hormones, or anything else that contributes to human thought and behavior. This amazing computer is likely no different than Google's object-recognizer.

Eventually, perhaps in 20-50 years, this will become closer to a realtime simulation. Even then, supercomputers would be the only ones able to even run part of these.

If/when we do get a true realtime simulation of all synapses in a human brain, there's zero guarantee it'll be anything more than that. Likely, it won't be able to do anything modern AIs can't. Consciousness is the je nai se quoi of biology, and I guarantee it requires more than just synapses (i.e. hormones, wants, etc.).