Eric M Russell (9/23/2016)
webrunner (9/23/2016)
Eric M Russell (9/23/2016)
Yet Another DBA (9/23/2016)
It is narrow-AI and not general-AI (which is what WATSON is trying to be).If we have problems with developers running feral in the database how many would want a megalomaniac AI paperclip to change the database?
Of course there are those who blindly follow all missing indexes and blindly follow all settings suggestions without testing 🙁 So would a piece of software be any worse?
:laugh:
I see more potential for practical application of narrow specialized AI than general WATSON style AI. For example, imagine in the aftermarth of earthquake we could have a ant-like swarm of AI nanites crawling across a collapsed building looking for survivors. It would be impossible to remote control a swarm of 10,000 drones, so each individual "ant" would possess enough AI for participating in a coordinated search pattern or alternately to branch off and go exploring on it's own.
That's a fantastic idea!
To keep things simple and cost effective, each nanite would consist of nothing but power supply, servomechanisms, camera, and a wireless network receiver and transmitter. The actual AI component of the swarm and the storage system for recording video feeds would be centralized in a portable suitcase sized computer connected to the internet. In the event that an individual nanite locates a victim, a human operator could initiate a manual override for it's system from any remote location.
Each individual nanite would possess it's own AI in the form of a separate managed process running on the central computer which would constantly be sending instructions to the nanite and receiving back a video stream and other sensory data. We're probably only a few years away from implementing something exactly like this.
After the search effort has completed, there would be no need to gather up all the nanites, they would basically become inert dust after a few days when their power systems drain. The ability to shut the entire swarm down centrally, the lack of an on-board AI, and a limited power supply would prevent the possibility of a rogue swarm taking over the city.
I see this type of AI application being so much more practical than WATSON.
Actually there is a Professor of Robots in the USA that argues that cheap smartphones attached to small drones that act as a swarm is the way to go. They can map out buildings and scan for issues far faster than fire rescue teams without the risks. Seeing what small drones can do is impressive. The issue I can see is that a small drone is going to be a lot faster than nanites outside, just due to headwinds and what power they can output.