In the last year Machine Learning, AI (Artificial Intelligence), Deep Learning and Data Science have become the new buzz words in information technology. Tech companies are now investing billions in the development of these new technologies and are racing to be able to boast the latest advance in image recognition or machines that can operate autonomously.

More and more consumer products are making use of Artificial Intelligence.

These products range from for your latest smartphone to your home thermostat or electric toothbrush. Consumers adopt these new technologies and products quickly, but companies seem far more hesitant and slower adopting products which contain AI.

Why are some companies hesitant adopting products containing Artificial Intelligence?

One of the reasons why companies are slower adopting these products is a gap in knowledge. Typically staff of an IT department have heard of Artificial Intelligence, and are more knowledgeable about it than people who are not working in IT, but they don’t know it as well as the tried and tested technologies they are currently working with. There is also a fear that products, which use AI, are going to make unexpected decisions, and that there is no way of finding out why a particular decision was made.

Making sense of AI, Machine Learning and other buzz words

Artificial Intelligence is an umbrella term for the field of Computer Science which deals with providing machines the ability to perform rational tasks such as Language Recognition, Automation, Image Recognition, and many others. The field of Artificial Intelligence includes:

  • Machine Learning is a subset of AI which is data oriented and deals with prediction. If you hear about techniques like SVM, Bayes and Decision Trees you can be confident that you are talking about machine learning.
  • Neural Networks are also a subset of AI which has been extremely popular over the last years. Neural Network algorithms mimic the Neurons in a human brain and are most commonly used in voice and image processing. The technique of applying Neural Networks is sometimes referred to as Deep Learning.
  • Data Science is often used in conjunction with Artificial Intelligence, however, it is not an algorithm, but a scientific field for extracting knowledge or gaining insights in data in various forms, either structured or unstructured.

What type of AI do we use for what application?

In general most Machine Learning techniques are used in a predictive capacity. Will the stock market rise or fall? Who will develop diabetes? What will the outcome be for a football match? Machine Learning is largely based on well-known statistic techniques and therefore the the results can be validated.

Neural Networks are typically used for non-predictive tasks such a face and voice recognition or autonomous driving cars. Large Neural Networks can be extremely complex, and unlike Machine Learning, it can be very challenging or nearly impossible to find out why a Neural Network made a particular decision. Currently, a lot of research is done in an effort to making the decision-making process of neural networks more transparent so that improvements can be made on current techniques.

Mind shift for adopting AI

Is it a bad thing that we don’t always know why a decision was made by a type of AI algorithm? The knee jerk reflex is the need to have control over a system and that we know exactly why a system did what it did. For the past decade, this way of thinking has been dominant in IT but it’s slowly changing to an approach where we don’t control the output of a system directly, but rather “train” the system like we do with a pet or apprentice. In the consumer market we can already see that we train our home thermostat to choose the best time to start heating your house, so for example, would it be so strange to let AI decide when it would be an optimal time to do maintenance on your IT environment?

If we are confident in the AI’s ability to be trained to a point where we trust its decision making, then it doesn’t really matter if we are not exactly sure why the AI made a choice, just that we know it made the right one.