Three Breakthroughs Unleashing AI

What this episode covers

Present day artificial intelligence started back in the 1950s, but why is it everywhere now? https://drpepermd.com/wp-content/uploads/2019/09/Three-Breakthroughs-Unleashing-AI-1.docx (#1 Three Breakthroughs Unleashing AI . transcript) There are three reasons AI is front and center. Number one is the ability for machines to do enormous amounts of computing very cheaply using something called GPUs and referred to as cheap parallel computation. Number two is massive amounts of data aptly called Big Data. And the third reason is better computer algorithms known as Deep Learning. Almost a decade ago a new kind of chip known as a GPU or graphics processing unit was developed for the video game world where millions of pixels had to be recalculated second to second, over and over again. Before long experts in the field realized GPUs could be used to run neural networks in parallel making computers more powerful in their computations. Networks running on GPUs enable companies such as Facebook to identify your friends in photos or for Netflix to create reliable recommendations for millions of subscribers. A second reason for the explosion in AI is due to the literal avalanche of collected data used by computers to learn and teach themselves. A computer needs to see thousands of examples before it can distinguish between, let’s say, a cat or a dog, or play a thousand games of chess before it can play the game at a proficient level. Twenty three hundred gigabytes of data are produced daily worldwide and it is astounding to realize that 90% of all the world’s data has been created in just the last 24 months. This has given us the big data computers need to train themselves and push AI forward. The third reason AI has taken off is due to something called Deep Learning. Deep Learning evolved from artificial neural networks which are collections of neurons based on the human brain except these are software based calculators that function in a similar manner to human neurons. Artificial neural networks are created when neurons are connected one to another and then organized into multiple layers. These deep layers of networks are why it’s called Deep Learning and has led to computers doing revolutionary things such as using speech to control devices like Alexa, and even analyze human emotion in customer reviews. It is this perfect storm of cheap parallel computation, Big Data, and Deep Learning that has created a 60 year in the making, overnight success of artificial intelligence. This is Dr. Peper from short and sweet AI. https://www.forbes.com/sites/bernardmarr/2017/04/25/the-complete-beginners-guide-to-artificial-intelligence/#202a441d4a83 (https://www.forbes.com/sites/bernardmarr/2017/04/25/the-complete-beginners-guide-to-artificial-intelligence/#202a441d4a83) https://www.wired.com/2014/10/future-of-artificial-intelligence/ (https://www.wired.com/2014/10/future-of-artificial-intelligence/)