Mary Jones said that Georgia schools are ahead of the curve in math, and she saw that students were studying at the same high level of math as she was.
She also played a game of rock-paper-scissors and found out that it is a game played all over the world,
and she used it in her classroom. She talked about the Chinese classroom system,
where students raise their hands silently to answer questions and problems,
and when the teacher recognizes them, the students can answer and ask questions.

How to train DeepMind’s machine learning models

In 2015, Nature published an article on how DeepMind trained a machine learning model to play arcade games.
Unlike dedicated chess supercomputers such as IBM’s Deep Blue, DeepMind’s goal was for the algorithm to master every game from scratch.
The algorithm was shown to accumulate information by playing games and to compete and win against highly skilled humans without the help of engineers. 

Loopholes in DeepMind algorithms

The algorithm quickly conquered Atari’s classic games,
but some of the classic games showed the algorithm failing to score a single point.
This was because the algorithm was designed to move on from one scoring opportunity to the next,
such as video pinball, where the score does not go up every time and the game is structured so that you don’t score until you collect the last item.