Expansion of artificial intelligence across most industries will create a major challenge for regulators, according to Geoffrey Hinton, a highly respected Google executive and pioneer in the booming field of deep learning.
Machine learning uses programs known as neural networks to replicate the way the human brain learns complex tasks, everything from recognizing pictures and sounds to understanding a language. Neural networks teach themselves to perform complex operations, taking the process out of the hands of developers.
This means that developers cannot explain exactly how they work and consequently cannot provide regulators with the necessary insights to be able to do exactly what they normally do, which is to regulate them.
All that is required for machine learning is a large amount of information on the right answer to any specific question, and the big neural networks can do the rest. Healthcare, oil and gas production, transportation and financial services have already adopted machine learning and artificial intelligence into daily operations.
Regulators in charge of most of those industries have a long way to go to catch up with how artificial intelligence affects their fields.
Yet, the impact of artificial intelligence and learning is profound and ever-growing. “Every single day we process the same amount of data as we did in the past in a month,†says Hinton, who has been at the forefront of AI development since its beginning.
It seems that the faster technology develops, the more regulators struggle to keep up. This is not only the case with artificial intelligence and deep learning but also with other areas of technology and IT which should be easier to regulate such as big data and virtualization. Maybe one solution would be to use artificial intelligence to improve regulation?