As soon as I entered Ted’s homepage, I saw a lecture called ‘The Danger of AI is weirder than you think’. And this title stimulated my curiosity. Perhaps the danger of AI, we think, is the story that AI will soon surpass human’s ability and that human will be ruled by AI in the future. But his title made me think that the presenter would introduce something different from what I know. As expected, AI researcher Janelle Shane warned of a slightly different kind of risk.
AI researcher Janelle Shane introduced the evidence to AI based on the research he had experimented with. As soon as I saw this picture, I realized that Janelle Shane’s danger would be nothing like I thought. This picture had nothing to do with AI’s dominance of mankind, and it had nothing to do with beyond human capabilities.
This experiment was done by Janelle Shane in collaboration with Kinder Middle School’s coders. And the subject of the experiment was to collect 1,600 flavors of ice cream and supply them to algorithms to observe what flavor AI produces.
Let’s take a look at the other experiments she’s talking about.
The experiment was done by AI, which used a robot to move robot parts from A to B. In fact, what we want from the robot is to pick up the parts one by one and then move the robot’s legs to move directly to B. In fact, robot scientists now use computer programs to use that method. However, AI is a self-study computer and when AI was given the task of solving the experiment, AI did the following.
Well… I think AI was never wrong. That is actually the right way to go. AI will solve exactly what we have asked technically. Can you say that AI is wrong just because he makes an ice cream called Pumpkin Trash Break and doesn’t use the path we want? I don’t think so.
AI researcher Janelle Shane said “AI can do what we want enough but we do the wrong thing by mistake. This problem is always a problem for AI. They can be destructive. But They don’t know ‘I am destructive’.”
And She advises that humans have a way to solve these problems. It is to develop communication skills with AI. By communicating with AI, we can reduce our mistakes if you know what AI can do or cannot do.
From this lecture, I realized that there were many prejudices and misunderstandings about AI. I’m sure that after watching this video, the bias against AI that you all know will be broken.(Here is the link: https://www.ted.com/talks/janelle_shane_the_danger_of_ai_is_weirder_than_you_think#t-598088)
This lecture made me think that it was useless to think that AI had control over humans. But that is not to ignore or denigrate AI. I saw the tremendous potential of AI in this lecture. We always wanted a specific answer, but AI didn’t. AI’s answer was one of countless possibilities. AI researcher Janelle Shane advised that communication with AI is necessary, but I thought we could apply AI to other areas to confirm the infinite possibilities of AI. When the human brain wants to know the field of no answer or many answers, it seems that our AI can provide the answer now.
AI has already given me a good answer. It was Ted’s ‘AI’ algorithm that introduced me to a really satisfying lecture!
T9Oliver219bm 