Artificial intelligence has sent shockwaves through the entire economic sphere and raised numerous ethical questions. For some, AI is like Eldorado, with potential for massive and unexpected increases in productivity. For others, the outlook is bleak, with AI posing significant threats to employment and even to humanity as a whole. Not counting the many startups for whom AI is at the core of their business structure, most companies are not yet equipped to understand what they can expect from AI, or to decide if and when they should consider integrating it.
Artificial Intelligence, or a prediction revolution
AI targets prediction. According to the authors, prediction is not “telling the future”. Essentially, it is using information that already exists to generate new information. This information concerns the future (what the temperature will be like tomorrow, for example) as much as the present. When Google Translate effectively transposes a text from one language to another, it is because its AI has “predicted” which words and syntax will correspond best with the intention of the author.
With the internet of things, neural networks, and other deep learning technologies, predictions are now much cheaper, faster and more reliable. In the past, taxi drivers needed many years of experience to be able to predict the best route to take, regardless of the traffic or weather conditions. Now thanks to tools like Waze, anyone can achieve the same level of efficiency without training.
The scope of AI is quite varied; image recognition, fraud detection, medical diagnosis, autonomous driving, and of course all parts of predictive analysis. It’s about recognizing patterns in available data in order to predict the likelihood of a particular outcome.
What will the role of people be in prediction?
Any prediction is subject to judgement, a complex process that takes into account both the desired objectives and consequences of each possible choice. It is only with judgement that prediction can lead to an intelligent decision and add value. For example, if an early warning device alerts you to a fraudulent transaction, it is only when you judge how to respond that this prediction will be of value to you.