The program was able to recognise 50 handwritten languages used in the Turing test – including Tibetan, Glagolitic, Sanskrit, and Gujarati – and proved to be even better than humans at recognising them, according to researchers. This is a huge leap forward for better Artificial Intelligence (AI) in the future.
Three researchers developed the program that can recognise handwritten drawings after viewing them a few times. The program also passed the basic Turing test.
Joshua Tenenbaum, co-author of the study and a professor in the Department of Brain and Cognitive Sciences at the Massachusetts Institute of Technology (MIT), said that this is the first machine system that is able to learn visual concepts in a similar way to human learners.
The Bayesian program learning framework (BPL) performed even better than people in some instances.
On one of the tests, both the Bayesian program and humans were shown a new character only once and then they were asked to pick the same character from a set of 20. The computer program had and error rate of 3.3%, compared with humans who had an error rate of 4.5 %.
The Bayesian program learning framework also performed well on the Turing test – which was developed by Alan Turing in 1950, a British pioneering computer scientist, to compare the thinking ability between humans and machines. People could not tell the difference between characters drawn by human participants or the computer.
Researchers say that other systems that rely on sign language, symbols, and written words may benefit from such performances. Dr. Turing predicted that one day people would speak of ‘thinking’ machines, without expecting to be contradicted by others.
To help the AI department develop, Elon Musk, founder of Tesla Motors and Space X, made two significant donations. In 2015, Musk gave 10$ million to the Future Life Institute (FLI), and at the end of this year he and other founders from YCombinator and PayPal donated 1$ billion to the new OpenAI foundation.
Future Life Institute is a volunteer-run research and organization in the Boston area that works to diminish existential risks that humanity may be facing, especially those coming from advanced artificial intelligence (AI). OpenAI is a non-profit artificial intelligence (AI) research company based in San Francisco, California. The OpenAI aims to develop friendly artificial intelligence (AI), which will benefit humanity.
Image Source: kurzweilai
Share this:
