Scientists and product developers at Microsoft are establishing a complementary AI approach called Machine Teaching, with an aim to make this technology even more powerful.
In the artificial intelligence (AI) scenario today, users feed a massive amount of data to the AI-powered systems and expect those systems to learn associations or find patterns on their own. For most of the common applications like simple text or image recognition, it works well. But the scenarios where AI is used today are growing, which is arising the need for Machine Teaching.
By machine teaching, Microsoft means that users and developers need to teach the AI systems not by bombarding them with a lot of data, but by breaking a problem into easier tasks (the way things are taught to children). What machine teaching aims to do is gain knowledge from people, instead of extracting knowledge from data alone.
The company used a very thoughtful example to explain this. If someone has to teach a person how to identify a table, they’ll start telling him that a table has four legs and a flat top. Further, if the person confuses the table with a chair, he’ll be taught that a chair has a back and a table doesn’t.
On the other hand, when it comes to making an AI algorithm learn what a table is, they’ll just feed a dataset with labeled pictures of tables, chairs, and lamps. Then they’ll expose the algorithm to several labeled examples of tables, which helps the algorithm to learn characteristics of the table.
“If you can teach something to another person, you should be able to teach it to a machine using language that is very close to how humans learn,” said Patrice Simard, Microsoft distinguished engineer who pioneered the company’s machine teaching work for Microsoft Research.
Also read: Impact of artificial intelligence on roles of business leaders: Microsoft research
“The reason machine teaching proves critical is because if you just use reinforcement learning naively and don’t give it any information on how to solve the problem, it’s going to explore randomly and will maybe hopefully — but frequently not ever — hit on a solution that works,” said Mark Hammond, Microsoft general manager for Business AI.
“It makes problems truly solvable whereas without machine teaching they aren’t.”
To help the AI systems and technology to tackle real-world problems more efficiently, Mark Hammond has developed a platform that uses machine teaching.
Read the detailed article by Microsoft here.
