I was checking my facts for a piece on AI and saw Elon Musk had posted this guys article on Twitter. Way better and more informed than I’m willing to do, so… Here you go.
There are a few points I disagree with here. First, computer/human integration will become tighter and more acceptable. Many people think people won’t want computer chips in their brain, that might not be the method of integration, obviously the method that the public accepts will be the method that will be commercialized and become cheap, but if you’re skeptical, imagine telling someone in the 60s about the prevalence of plastic surgery today. You might say, well it’s not the majority of the population, and I’d point out not everyone is obsessed with their body image, everyone is obsessed with the Internet. Combine that with a cheap surgery that you probably won’t even have to be put under for, and may even be an injection, (unlike most elective surgeries today) we will see mass adoption. What’s the point of this setup? There will not be an AI, there will be many AIs and we will BE the AI and the AI will be us.
Safety research is important, there will be accidents, but the research is being done, there will not be a revolt of the machines… So worry about the zombie apocalypse instead.
AI will seem intelligent in 10 years. Most people don’t talk about it, but stop and think about it, imagine Apples SIRI in 10 years combine with Watson (IBMs computer that beat the top Jeopardy players) in 10 years. You’re going to be able to ask your watch any information and it will be able to reply with the exact and correct information, it will have its own personality that may evolve, and hold extensive conversations. YouTube videos of people putting two phones together and see what conversations they have will be a thing. People will empathize with their phones like on the movie Her. Apple SIRI, Amazon Echo, OK Google, Microsoft Quortana. The makers of Siri are making a new one (Apple bought Siri) called VIV.
For computers to take over research, computers do not have to be as intelligent as a person. Many of the human brain neurons are devoted to emotions, hand eye coordination, and left over from evolution, etc… Sure it works great, most people are happy with their brain, but it’s not exactly ideally designed for logic and scientific research. In fact, all a computer needs to take over the research is being able to perform the tasks, simulations or robotics, a basic understanding of the problems and goals which is being done now, and intelligent path finding. This type of setup is already being used for network penetration tests (test hacking websites). Computer guided development and evolution of computer intelligence will take off in 20 years, I believe this is the knee of the curve for commercial intelligence and not truly human level intelligent computers. We don’t want a human computer, we just want a smart computer thats reasonably good and faking an understanding of emotion.