Why You Shouldn’t Fear Artificial Intelligence

2 Comments

1. Grant January 20, 2015

   I am surprised he didn’t bring up the fact that the bulk of stock trades on the stock market are now done by AI and that they “have” gone wrong from time to time, losing billions of dollars and causing shut downs. Their human overlords said the trades and the decision making that drove them were being done so fast and with such complexity that they (the overlords) no longer understood exactly why they did what they did.

   That example brings up a good point. The programing was so complex and required such a large team that no “one” human could fully understand how it functioned (not unlike any large software program). Although they may have had a blueprint in the early designs of the program, they didn’t/couldn’t account for the implications of competing AI all working at lightening speeds. It was developed in unchartered territory just like many upcoming AI programs will be. I could see that being one hole in AI development to watch for as AI based programs begin to effect more critical elements of society.

   The other gaping hole is the one mentioned which is that the AI programs are being built with the human shortcomings of its programmers consciously or unconsciously being built into it. Consciously in the case of AI programmers who “want” to build an AI program to do specific damage and unconsciously by programmers who may be unaware of their own personal biases. The latter is obviously a big rabbit hole to go down when it comes to defining normative or acceptable behavior on a Global basis.

   I do think it is important to be worried about these things early in the development cycle before the AI is designed to “learn how to learn”. We may already be too late that front!
   • HIGH T3CH January 22, 2015

     That’s a good example about stock trades. I’ll have to agree with you. It all seems like a natural evolution, perhaps AI is the pinnacle of evolution, I certainly hope not…