AI – from theory to improving the mobile experience

Down Arrow

Posted by Johan van der Steen on 29 Nov, 2017

Future Platforms Spotlight Series in September – Artificial Intelligence

Whether your company is ready or not, artificial intelligence (AI) is here. It’s no longer a science fiction fad. Embrace the technology and start preparing for how you can utilise it to your advantage.


What is artificial intelligence?

Artificial intelligence has been around for over 60 years. AI research goes back to the summer of 1956 – founded at a workshop at Dartmouth College. Even with this long history, there’s still some confusion about what the technology actually is and how it works. The best way to describe artificial intelligence is to compare it with an algorithm.

Algorithms and AI – key differences

An algorithm is a written set of rules that should be followed to achieve a desired outcome. A cake recipe is an algorithm. The recipe explains exactly what ingredients, measurements, temperature and timings are required. The algorithm simply follows the steps as described. Et voila – a delicious cake!

AI is slightly different – almost like a technical challenge from the Great British Bake Off. You’ll provide the artificial intelligence machine with the cake ingredients and then let it figure out the rest. It might not get it right first time, but through machine learning it will tweak its process and improve over time.

How does AI learn to get it right?

Take a look at the image below:

Artificial Intelligence - inputs and outputs
At the start, you provide input into your AI machine with multiple sets of data. The machine then works through this to provide you with the desired output, by taking your data through the different layers of nodes (represented by the circles).

AI machines process data through these nodes by using mathematical formulas – like regressions. They follows various parameters that are set in place from the beginning. Ultimately, the machine will not be able to tell when it’s reached its end destination. Instead, it gives a total score to suggest how confident it is in reaching the intended outcome. In the image above, examples of an output which artificial intelligence can work in, are price projection (certainties of price), image tagging (identifying objects) and word identification (understanding the words you say/write).

A simple explanation of how AI works

Imagine a machine playing the video game, Super Mario. A machine can play through the course, jumping through various obstacles until Mario dies. It restarts from the beginning but takes care to avoid the final step before its death. It takes an alternative path. If successful, it continues – if not, it starts again. Here, the machine is provided with the input data (pixels on screen) and works through the layers until it reaches the end goal (staying alive). It keeps doing this until the character gets to the end destination, completing the course without dying. This is the ‘train and learn’ approach.

In the rest of talk, Olivier further explains the technology and explores the different business cases where artificial intelligence in mobile could be of benefit to organisations in the future.

If you would like to view the full video of Olivier’s talk or if you are interested in attending the next Spotlight Series event, please send an email to eddie.guevarra@futureplatforms.com.

Get in touch with Future Platforms