This is the topic of the day. AI (Artificial Intelligence) this, AI that, everything seems to circle around AI now. The subject is fascinating and is rich with promise. But, what is AI? Everyone knows and has, more or less, been exposed to natural stupidity. Many talks about artificial intelligence without knowing what is behind these two words. Often, a lot of fantasies gets spread.
In Stanley Kubrick’s film, 2001, A Space Odyssey, the spaceship’s computer, HAL 9000, is a form of artificial intelligence. He’s actually a crew member, almost like the others. Astronauts talk to him, he answers, reasons, and takes care of the whole ship. Until he refuses to open an airlock to a crew member. He understands that the crew wants to disconnect it. It is the singularity evoked by the late Stephen Hawking and Elon Musk. The computer becomes smarter than humans and decides to cut the latter. It does not see humans with respect but as its enemy. And this is how in the collective mind, the intelligent computer was built and is always present.
If an AI that reads on our lips, analyzes our feelings in real time, and answers our questions with brio, will exist, we are not there yet. Yet, AI is present in our daily lives. It’s everywhere behind websites and applications that we use daily. It analyzes our actions, follows our digital wanderings and builds behavioral models. But what is it?
For the moment, Ai is “narrow”, meaning that it focuses on a given task. For example, analyzing faces in a picture, recognizing the intonation of a voice, or reading a text and understanding its meaning. It knows how to use the engine and controls of a car or airplane according to various constraints. It can also make recommendations based on choices and historical events. Generally speaking, an AI is often better than a human being for logical and repetitive tasks that do not involve creativity. That’s where the shoe pinches today. If an AI is capable of imitating Rembrandt to perfection, it only knows that. This same IA will not know how to compose music, write a poem or sculpt marble. These are “narrow” processes, not devoid of interest.
So where are those artificial intelligences? In the research labs, of course, and also in the data centers of the cloud. From Microsoft to Alibaba to IBM, Google, Facebook, and Amazon (more information soon in my forthcoming book). The usual suspects!
We can group narrow AIs usable by everyone today under 7 distinct categories:
- Fixed image analysis
- Video analysis
- Speech recognition and synthesis – UPDATE: Google Duplex is demonstrating pretty impressive speech capabilities, read more here.
- Conversational Robots
Cloud providers offer each of those services. They can be aggregated to provide complex services. For example, linking image recognition and speech synthesis will allow a scene to be described to a blind person. Analyzing an image and passing the result to a conversational robot (chatbot) will allow estimating the damage of a leak. Talk to a computer, and the latter interprets and translates on-the-fly. These services are integrated into the applications of our smartphones and computers.
They will also provide AI on the edges of the cloud, instead of running them from datacenters. In the years to come, with the deployment of fiber, 5G and satellite broadband, and the increase of computing power through quantum computing, these services will be perceived as natural and indispensable. They’re going to transform our world. This is one of the reasons Klaus Schwab, the founder of the Davos Forum, talks about “Fourth Industrial Revolution”.