a lo-fi blog of modern Spokane life

Home - Uncategorized - Perhaps a Way to Solve All Our Worries About AI (Artificial Intelligence)

Perhaps a Way to Solve All Our Worries About AI (Artificial Intelligence)

Posted on October 19, 2017 in Uncategorized

So I was reading an article about the Google/Deep Mind AI that has learned to play the game of Go. One of the things that stood out was this photo of the matches where the AI first beat a human at the game of Go. When you look at it, they are all humans. Similar to the old days of Watson beating masters of chess, there is a human standing in for the computer to actually move the pieces.

This would seem to offer an obvious response to the concerns about AI taking over or having the potential to damage humans in the future. We should identify a few basic tasks such as driving, operating trains, transport trucks and the like where AI and autonomous actions by the robots would be helpful. But we should then define a number of areas where AI can not interface with humans without another human intermediary. This would preserve some of the jobs that may be lost (humans as the butlers to robots? haha) but it would prevent robots from taking over. The risk of AI taking actions with vehicles like cars and trucks that would mimic terrorists (crashing into a crowd, etc.) is real, but likely can be limited with some serious protocols and tracking features.

This is kind of where Elon Musk has said that we need to be very worried about AI. The self driving car is a wonderful dream. The safety gained by not having human drivers can likely be obtained with other innovations in transportation that does not require autonomous driving. So this is a specious argument and a red herring when you think about it. I mean we could have helping features that would assist in night driving or other areas of common accidents, and we could also change our laws to enforce very strict guidelines on not having cars close to each other. In that scenario the reason for the lack of safety is because it is not worth the cost of police enforcement or change in our behavior. If the reasons for robots to take over driving is that we do not want to pay for the costs of making ourselves safe seems like a stupid line of questions. Also, simple actions like having breath testing or sleepy tests in cars before they start would also alleviate a lot of trouble.

But here in this little photo is the key to future AI. AI can live on a circut board. AI can be in the cloud. It can be in the hard drive. But when AI interacts with the world, it needs a human butler. This is the key.

Leave a Reply

Your email address will not be published. Required fields are marked *