Let's Talk: Learning Experiences For Machines
Updated: Sep 30, 2019
Understanding how Artificial Intelligence (AI) works will lay the groundwork for understanding how Artificial Intelligence can effect the world of learning. Machines powered by Artificial Intelligence do not acquire, process and present information in the traditional method of loading data with pre-defined data structures, processing data with predefined algorithms and presenting the processed information in a pre-defined format. With Artificial Intelligence, machines learn through experiential learning and begin to make choices, form strategies, and transform the world around them. New learning experiences and learning theory for machines will evolve and MIT has already started doing it.
In my own voice - 2:00 video:
Evidence for this theory lies in the findings of an experiment recently published by MIT in the article, AI learned to use tools cooperation after hide and seek games. The artificial intelligence learning experiment sets players using artificial intelligence in game of hide and seek. The hiders or seekers are rewarded or punished based on the performance needs to keep the game active and relevant. The AI players competed in the game 380 million times before beginning to exhibit evidence of strategy and then 500 million times before beginning to learn to collaborate with other players who share common goals. The constant feedback through experience and reflection shows evidence that a machine can learn through experience. A learning design professional could make these learning experiences more effective, efficient and ethical.
Artificial Intelligence in a Nutshell:
The meta-cognitive programming gives the machine the choice on how to receive, process and present information. It is much like playing the game called Tetris. You are given an option and see options coming and make choices based on different levels of memory instead of following a predefined step by step instruction. The information coming in is not stored long term or short term unless the machine determines it should be. Therefore the information is curated based on the experiential learning that occurs (much like humans).
Here are the actions for handling information with us :
1. Determines stored in short term memory (easy access)
2. Determines stored in long term memory (longer, but not always easier access)
3. How to use the information - determined connections or wisdom
4. Purge the information.
What to think about?
This is only the beginning of using experiential learning design for both humans and machine learning objectives. And yes, a multitude of complexities such as these could exist for the following:
1. How do humans interact with machines while learning together?
2. Can a machine learn empathy while learning with humans and developing strategy together?
3. Will a machine develop different cognitive stacks developing its intellectual capabilities from different learning experiences.
4. How will this change, advance or use learning theory as we know it today? A new variation of Blooms Taxonomy perhaps?
To get the oral history of Blooms Taxonomy from Dr. Lorin Anderson watch the Off The Cuff videos produced by . Dr. Anderson studied at the University of Chicago where Blooms Taxonomy was born and received his degrees from the money raised from the sale original Blooms books.