However, these AI algorithms cannot explain thought processes Behind their decisions. a Computer That owner protein folding And it also tells researchers that more information about the laws of biology is more useful than a computer that folds proteins without explanation.
Hence, AI researchers like me Now our efforts are changing towards developing AI algorithms that can explain themselves in a way that humans can understand. If we can do this, I believe AI will be able to uncover and teach new facts about the world that have not yet been discovered, leading to new innovations.
Learning from experience
An area of AI, Reinforcement learning is called, How computers can learn from their own experiences. In reinforcement learning, AI discovers the world by receiving positive or negative feedback based on its actions.
This approach has led to algorithms that are independently learned Play chess on a supernatural level And prove Mathematical theorem Without any human guidance. As my work An AI researcher, I use reinforcement learning to create AI algorithms that learn how Solve riddles like rubik cube.
Through reinforcement learning, AIs are learning to solve problems independently that even humans struggle to know. This got me and many other researchers thinking about what AI can learn and more about what humans can learn from AI. A computer that can solve a Rubik’s cube needs to be able to teach people how to solve it.
Peek into the black box
Unfortunately, the minds of the supernatural AIIMS are currently out of reach for us humans. AI makes awesome teachers and in the computer science world we say “black box“
A black-box AI spits a solution only without giving a reason for its solution. Computer scientists are trying to It took decades to open this black box, And recent research has shown that many AI algorithms actually think in ways that are similar to humans. For example, a computer trained to recognize animals will learn about different types of eyes and ears and put this information together To correctly identify the animal.
The attempt to open a black box is called Explainable A.I.. My research group The AI Institute at the University of South Carolina is interested in developing interpretive AI. To accomplish this, we work heavily with the Rubik’s Cube.
Rubix Cube is basically a Pathfinding problemFind a way from point A – a fried Rubik’s Cube – a Rubik’s cube solving point B. Other pathfinding problems include navigation, theorem proving, and chemical synthesis.
My lab has set up a website where anyone can see how our AI algorithm fixes rubic cube; However, a person will have to work hard to know how to solve the cube from this website. This is because the computer cannot tell you the reasoning behind its solution.
Rubix cube solutions can be broken into some generalized form Step– The first step may be, for example, to make a cross while the second step may be to place the corner pieces. While Rubik’s Cube itself has possible combinations of 10 to 19th power, a generalized step-by-step guide is very easy to remember and applicable in many different scenarios.
Breaking and accepting a problem in stages is often the default way in which people explain things to each other. Rubik’s cube naturally fits into this step-by-step framework, which gives us the opportunity to open our black box Algorithm more easily. Creating AI algorithms with such capability can allow people to collaborate with AI and break many complex problems into easy-to-understand steps.
Collaboration leads to innovation
Our process begins with using our own intuition to define a step-by-step plan thought to potentially solve a complex problem. The algorithm then looks at each individual step and gives feedback about which steps are possible, which are impossible and the plan can be improved. The human then refines the initial plan using advice from the AI, and the process repeats until the problem is resolved. The hope is that individuals and AI will eventually convert into a kind of mutual understanding.
Currently, our algorithm is able to consider a human plan to solve the rubic cube, suggests scheme improvements, identifies schemes that do not work and options that are available. When doing so, it gives feedback that leads to a step-by-step plan to solve a Rubik’s cube that a person can understand. Our team’s next step is to build an intuitive interface that will allow our algorithm to teach people how to solve a Rubik’s Cube. Our hope is to generalize this approach to pioneering problems.
People are unmatched by any AI, but machines are far superior in terms of their computational power and algorithm rigidity. Between man and machine it uses both front and back strengths. I believe that this type of collaboration will shed light on all unresolved problems ranging from chemistry to mathematics, which may lead to new solutions, intuition and innovation, otherwise out of reach.
Quotes: How Explanatory Artificial Intelligence Can Help Humans New (2021, 13 January) Retrieved 13 January 2021 from https://naveenbharat.com/news/2021-01-artific-intelligence-humans.html
This document is subject to copyright. No part may be reproduced without written permission, except for any fair which serves for the purpose of personal study or research. The content is provided for information purposes only.