Breaking down the concept of Artificial Intelligence

If you want to read the spanish version click here

First things first, what’s intelligence?

According to Marvin Minsky (one of the fathers of AI), intelligence is the capability to resolve issues or problems that have not yet been solved. Therefore, it would not be wrong to assume that Artificial Intelligence could be defined in the same way when considering that the tasks needed to resolve those issues or problems would be undertaken by a machine.

Did you know there are different types of Artificial Intelligence?

Nowadays, it is common to relate AI to innovative processes, resulting in daily solutions already based on this set of technologies to be deemed as artificial intelligence. Examples of this would be robot vacuums and the GPS systems of our mobile phones. However, these examples perfectly fit Minsky´s definition mentioned above as these machines have been given the capability to resolve an issue that would otherwise require human intelligence to be resolved. We would qualify these day-to-day examples of AI under the category of Weak Artificial Intelligence.

AI enjoys an array of possible implementations and sitting the furthest from the just explained Weak AI, we would find Strong Artificial Intelligence, which is the type of AI that would allow a machine to behave like a human. However, this vision is far from being incorporated into industrial applications as, despite the current advances of psychology, neuroscience or philosophy, there is not an exact framework that defines the different human behaviours as such and therefore resulting in Strong AI still being a long way to go to be incorporated into computer systems.

So, what do we at Lurtis Rules class as Artificial Intelligence?

We consider a machine to be intelligent if, taking into account certain criteria, restrictions and evaluations, it is capable of operating most efficiently to obtain coherent results with complete autonomy. In other words, without anyone having to tell said machine how to tackle the task at hand in order to reach objectives efficiently.

Let’s not get terms mixed up...

Now that we have briefly explained what Artificial Intelligence is (believe us when we say briefly, we can talk about it for days on end!), some other terms in the field tend to generate confusion, such as Machine Learning, Deep Learning and Data Science.

Have you ever wondered which one is better: Artificial Intelligence or Machine Learning? If so, this is an excellent question to have. The simplest of answers is that Machine Learning is a subdiscipline of AI and therefore, they are not really up for comparison.

Machine Learning is a discipline based on data collection and the recognition of patterns that then allow them to create systems capable of automatically associate, segment and/or classify said data. Hence, the intelligent feature the computer mimics is learning, particularly learning from samples, being these samples either data or scenarios that the computer proposes and can evaluate. For example, suppose we have a data set of materials with information of their own properties and other characteristics. In that case, Machine Learning could be used to search for patterns and categorise this information in a way that it would allow for the findings to be utilised to perform future simulations more efficiently.

A subdiscipline of Machine Learning, Deep Learning, is also a popular concept. To explain this, imagine a Black Box that receives certain information and can get the patterns, classify information… but it can’t tell what is actually happening. This subdiscipline is based on Neural Networks with multiple layers which have been proven very effective in problems like image recognition, voice processing, but with a lack of explanatory power which make them applicable in certain fields but not in all of them.

Another term that causes confusion is Data Science, a discipline which benefits start from its capability to automatically extract data, categorise them… to the analysis of information to then be able to make coherent decisions. In order to this happen, Data Science involves tools and techniques related to AI, so that the machine looking after said data could work on its own and consequently learn. Also involves other techniques that come from data management, data visualisation/interaction, or other math-related modelling or simulation.

When the amount of data reaches over millions or comes from a great variety of sources, it can be considered Big Data, that stops being just an information manager and turns into a provider or feeds that data to other techniques or settings to manage data storage.

Allow us, before we conclude this post, to refresh the concepts mentioned above in the following scheme.

This is it for now! We hope you enjoyed this blog and that the above concepts are a bit cleaner. Do stay tuned for what’s to come as in our next blog post we will explain how to approach implementing Artificial Intelligence into your business or project (without dying on the attempt!).

Should you have any queries, please do not hesitate to get in touch at info@lurtis.com. Don’t forget to subscribe to our Newsletter. See you next time!

Lorena Cruz - Business Development Lead

Subscribe to our newsletter and receive all our latest updates