A summary of an afternoon spent on AI and what to expect soon.

After a very interesting afternoon spent at Holberton school in the company of some of our mentors, Louis Monier and Gregory Renard, we learned a lot about the history of and the developments in Deep Learning systems. Artificial intelligence (AI) has been changing our lives for decades, and has never felt more present than today. It seems as not a week passes by without another AI system performing different tasks to help bring a change in our everyday life.

2015 was the tipping point of the adoption of the Internet, digital medical devices, blockchain, gene editing, drones, and solar energy. 2016 will be the beginning of an even bigger revolution, that could change the way we live. No one knows for sure how the future of AI will expand. This future could be promising, by making all our dreams come true, but also dangerous and destroy society and the world as we know it.

There have been major advances in “deep learning” neural networks, which learn by ingesting large amounts of data as we can see with IBM’s A.I. system, Watson, that learned everything from cooking, to finance, to medicine; and Facebook, Google, and Microsoft have also made great improvements by developing face recognition and human-like speech systems. These type of system has almost reached human capability, and IBM’s Watson can diagnose certain cancers better than any human doctor can.

Here is how it works and what to expect soon:

Some of the main points highlighted by Louis Monier and Gregory Renard on deep learning were neural networks, big data and the right algorithm.

Thinking is an inherently parallel process, billions of neurons firing at the same time to create synchronous waves of cortical computation. To build a neural network (the primary architecture of deep learning systems) it also requires different processes to take place simultaneously. Each node of the neural network imitates a neuron in the brain, mutually interacting with its neighbors to make sense of the signals it receives. To recognize a spoken word, a program must be able to hear all the phonemes in relation to one another; to identify an image, it needs to see the context of the pixels around it.

Every intelligence has to be taught, and the best way of doing so is to give it as much data and practice as possible to make it as efficient as possible.

By organizing neural nodes into stacked layers, each layer is found to trigger a pattern and emit a result leading to another layer. It can take many millions of these nodes, stacked up to 15 levels high, to recognize a human face.

From self-driving cars, robots performing tedious tasks, to the development of Virtual Reality, and having everything in your household connected is what to expect soon.

If all this is happening today, you can imagine the excitement in the world of tomorrow!