This is what Google wants to prove the human world with the Magenta project. Revealed in 2016 during a music & technology festival in the USA, the Magenta project’s goal is to improve our machines’ creativity. By using artificial intelligence (AI), Google wants to allow computers to create music, videos or pieces of art.
Let’s go a few years back: in 1997, Deep Blue, IBM’s super computer, beats the chess master Garry Kasparov. This is the first step: humanity “bows” to the machine. Less than 20 years later, Alpha Go unsettles the Go game world champion (Lee Sedol) during a game that attracts the eye of specialists all around the world. The gap between the win in 1997 and the one in 2016 is significant: in chess, the number of possible combinations is around 10^120… against 10^600 for the Go game. (For those of you who don’t like math: imagine the number 1, and add 600 zeros!)
One of the reasons for this success is deep learning: the ability for a machine to learn, but mostly to learn on the basis of what it learned before. The purpose of this article is not to detail the complex processes linked to the profound neuron networks, but rather to make a situational analysis that will allow you to understand how AI will change our day-to-day life.
From Google to IBM and Apple, the number of companies who want to capitalize on the first progress is huge. We can mention the French company Snips, or the online leader Amazon. Facebook is currently developing partnerships with European Universities to make the necessary resources available to allow progress in research, in the medical field for example.
Whereas human intelligence is thought to have appeared 3 million years ago, AI is obviously more recent. Allan Turing is one of the AI pioneers, but we have to wait until the 70’s and the emergence of algorithms to start seeing what is possible. Investments related to AI explode at the beginning of the 21st century. Fundraisings multiply, in the Silicon Valley but also in China: a 13 billion euro investment plan was announced in May 2016, and what we create in terms of AI is not even “wahou” yet!
More and more companies allocate a part of their R&D budget to AI. It’s hard to give numbers, because they keep it a secret. However we know today that programs are moving forward, and broaden the scope of possibilities:
To go a bit deeper and understand how different these learnings are, find which of the above programs match the following:
Today, AI is not generic, it only applies to specific needs. It’s only efficient in delimited and monitored frameworks.
AI is quite convenient (and pretty awesome). But we hear more and more specialists (Bill Gates for instance) take firm positions on the risks of “super-intelligence”. They consider these forms of intelligence are a threat to human-kind. Why? Well because it is very hard to have enough visibility to understand when and how AI will be relevant and efficient in non-specified, non-monitored, non-framed contexts.
Siri’s founders have developed a personal assistant – Viv – able to understand “natural language” like any other program. Which makes it able to evolve in a development with no boundaries at all..
Scary for some, amazing for other, one thing is sure: we aren’t done talking about AI!