Machine Learning and Artificial Intelligence – Promises of the future

Machine learning (ML) and artificial intelligence (AI), though different fields are both concerned with enabling machines to become more useful overtime. Learning machines do this automatically and independently  and provide this usefulness to humans in a more natural way.

In this article I will briefly illustrate the basic difference between ML and AI. I will also  provide some examples of how ML and AI are used now and what we can expect from them in the future.

Distinction Between Machine Learning and Artificial Intelligence

Machine Learning

ML is concerned with the development of algorithms that give computers the ability to ‘learn’ to perform a task without being explicitly programmed to do it. Examples include an online store ‘learning’ what products you may be interested in or what videos you might like. The computer does this by observing your shopping or watching habits over time and recommending similar products or videos.

ML can also be used to allow a robot to learn to navigate in an environment with telling it how it is to react to every possible ways it can detect an obstacle. This is the way it is traditionally done and requires the programmer to account for every possible way the robot could hit an object and then tell the robot how to avoid it. This can work but can also fail miserably in an unstructured, constantly changing environment, like the real world.

The following videos show two applications of ML in action. The learning process is sometimes slow, so it can be tedious to watch, be patient.

Robot Learning to Avoid Obstacles

Computer Learning to Play a Video Game


Artificial Intelligence

AI, though it may use machine learning techniques, is more concerned with developing machines that simulate or mimic humans. AI is aimed at mimicking the way we  interact  and think when solving problems. The best AI will be able to easily make a human think he or she is interacting with another human. This is the basis of the Turing Test which is a long standing goal of AI researchers.

AI advances will lead to more productive interactions between computers and humans. This will allow greater efficiency and productivity as some everyday processes can be automated without being cold. Such as customer support for example, or banking or even a fitness instructor.

The following video gives a definition of AI and recaps its original definition and goals. This gives us an idea what the original AI scientist thought and where we are today.

This next video tells us what the United States emerging technology development agency (in practice the Tech developed is way out there, thick scifi) thinks of AI now and where it’s going. This agency is Defense Advanced Research Projects Agency  or DARPA, learn more about them in this Wikipedia article.

Machine Learning and Artificial Intelligence in the Present

ML and AI already play a significant role in our daily lives now and advances promise greater benefits in the future. In our daily online life, ML and AI leads to better and more relevant recommendations of products on Amazon or movies on Netflix.

They lead to easier interaction with with electronic devices such as the Amazon Echo or the Google Home where they allow the devices to understand our natural speech, using so called natural language processing, and respond appropriately.

Even in the financial world ML and AI allow for better trades and investments with higher returns by analyzing and learning from the vast amounts of past data on stock prices. In banks combing through this historic data allows ML to learn and predict. This is used to help detect fraudulent transactions, such as someone using your credit card without your permission.

In some of these applications the power of ML and AI often goes unnoticed and this can be viewed as a testament to the natural and indispensable usefulness that they provide in daily life.

Promises for the Future

For the future it is expected that ML and AI will provide more exciting applications that promises greater benefits to us in terms of safety, productivity, sustainability and economics. ML and AI are already being successfully deployed commercially in cars and allows the cars to automatically and independently perform common tasks that a driver would perform.

GoogleSlefDrivingCar Machine Learning

This Self Driving car technology is already proving itself to be a safer option to human drivers. Improvement in safety is already being proven with the Tesla autopilot able to achieve a 40% crash reduction, see details here. Self Driving car technology is already recognized as the inevitable near term evolution of automotive technology. This is easily seen by looking at the growing list of big car makers working on this technology.

MRI Machine Learning

ML and AI are also aiding professionals in performing their job at a higher level. ML and AI help doctors to diagnose diseases earlier and with greater accuracy, it also allows them to understand and predict risk factors for certain diseases. AI is able to detect diseases that doctors normally would miss, it is therefore predicted to replace doctors (and so many other professions) in the future.

In the legal profession ML and AI helps lawyers to analyze huge amounts of past cases to detect patterns. This could aid in developing strategies and a possible winning defense. AI can help to formulate new strategies and ML can evaluate the success of past strategies, They could potentially even also help lawyers determine which judges may be sympathetic to their case.  In this application, ML and AI are used to analyze past written judgements made by a specific judge. Can ML and AI replace lawyers?

In Closing

ML and AI will become more advanced as time goes on and this will lead to machines becoming more intelligent. As with all technology this can have good and bad uses. Therefore it will be up to society to ensure that the intelligent machines are of benefit. This will require society to understand what ML and AI are and how they function. In this way we can make good decisions regarding how they should be used and if we should regulate them.

In the future I will be writing articles on some of the technical details of machine learning and artificial intelligence. These will include design, implementation and testing details. I will try to keep the math and programming manageable. But be warned (or excited) that it can only be simplified so much and no more. You can Follow me on social media so you will know when these post are out.