Date of publication: 2017-08-27 14:16
In human intelligence and culture, uncertainty and probability are called 8775 human potential 8776 . We humans are UNSUCCESSFUL in trying to put positive limits on that uncertainty because we 8767 re using folkways and mores from RELIGION to make judgments and choices. Laws/commands from that tradition are not working because taboos are routed to the same unconscious space as that of unmanifested potential ., there are programming models to AVOID, including that of the Illuminati religion (to accept 8775 temporary 8776 evil to bring greater good [for themselves]).
The New York Times | 8775 The future is now? Pretty soon, at least futurist Ray Kurzweil sees a revolution fueled by information technology 8776
Fox News | 8775 Elon Musk says we are summoning the demon with artificial intelligence 8776
Fox News | 8775 Could artificial intelligence be humanity 8767 s biggest threat 8776 : video
Ultimately, the most important approach we can take to keep AI safe is to work on our human governance and social institutions. We are already a human machine civilization.
Everything about the evolution of AGI hangs on how long it takes to reach that understanding for itself, how long before it makes those abstractions?
If it gets there in seconds or minutes, we will probably survive.
If it takes hours or days then the survival probabilities for humanity start to seriously degrade.
Overall, I think mathematical AI has a better best case but also a worse worst case than neuromorphic. If you really want goal preservation and think goal drift would make the future worthless, you might lean more towards mathematical AI because it's more likely to perfect goal preservation. But I probably care less about goal preservation and more about avoiding terrible outcomes.
I'm skeptical of #6, though I suppose if the AI is very alien, these kinds of unknown unknowns become more plausible. #7 is an interesting point. It seems like a pretty good way to spread yourself as an AI is to become a useful software product that lots of people want to install, ., to sell your services on the world market, as Hall said. Of course, once that's done, perhaps the AI could find a way to take over the world. Maybe it could silently quash competitor AI projects. Maybe it could hack into computers worldwide via the Internet and Internet of Things, as the AI did in the Delete series. Maybe it could devise a way to convince humans to give it access to sensitive control systems, as Skynet did in Terminator 8 .
Genetic engineering will increasing allow parents to screen and tune the traits of their children. Sale of gametes and surrogacy services will become more widely accepted. Cloning will mainly be used for reasons of sentimentality and reproductive difficulty. Eugenics will never be a mandatory social policy, but widespread voluntary genetic engineering will have a similar effect.
what seems consistently left out of discussions on either a harmful or helpful AI, as in a human level and self aware AI, is the likelihood that by the time computing and robotics reaches sentience, humanity will be directly integrating machine intelligence with our own. we already are on this path of course, I store all my phone numbers, movies to watch, records to listen to and so on externally on the cloud. mr kurzweil mentions this in terms of informational access, with the smartphone vs 85 8797 s president example.
Time and energy might be better spent trying to prevent the evil human people that currently commit horrific acts. I 8767 m hopeful that a sufficiently programmed AI might eventually help resolve some of the conflict and needless suffering that many humans experience today.
Third – unless the human kernel directed it, the AI wouldn’t be remotely interested. I would venture that the human kernel wouldn’t be enthusiastic either unless ordered by their sponsors.
Privacy. Privacy will be affected by technology in two ways. First, information processing technology will allow businesses to track customer interactions to arbitrary levels of detail. Customers who consider this a problem will seek businesses offering privacy guarantees, or will patronize intermediation services that offer to protect their identity. Public-key cryptography will always allow secure and private communication and e-commerce even if network traffic can be intercepted.