Developments such as Artificial Intelligence (AI), Machine Learning (ML) and Deep Learning (DL), have begun to impact the solutions of the audiovisual industry.
Andrea Mereghetti*
We have already left in the past one of the most controversial years in the history of humanity, we have reached a level of social stress that I believe is comparable only to past events such as the Red Brigades in Italy, the Cold War (USA and the former USSR) and the threat of nuclear war, just under 30 years ago. The likes of Donald Trump, Kim Jong-un, and other populists around the world have defined and are defining our future and the way we perceive what is to come.
Alan Touring, mathematician and brilliant mind, the father of the Science of Computational Theory and Artificial Intelligence (AI), as well as "Codebreaker" in the Second World War, received official apologies in 2013 from the British Government of Gordon Brown, after his suicide at the young age of 41, for having been discriminated against and humiliated for his homosexuality.
Yours is the "Touring Test", which for decades was indicated as one of the tests par excellence to be able to recognize if you were interacting with a machine or with a person, is it still valid? What are the tools available to define or realize differences? Well, there isn't.
These technologies (AI) are defining our lives and little or nothing is being done to protect us. In 2017, taking the laws1 of robotics defined in 1942 by Isaac Asimov (famous writer of the last century, author of the Foundation Trilogy, I Robot, and many other science fiction stories) the European Union made a first attempt to reduce a possible impact on society.
Elon Musk, one of the visionaries of this century, on different occasions stressed that it is time to regulate Artificial Intelligence (AI) before it is too late, it is surprising and at the same time fascinating how SpaceX (one of his companies, in addition to Tesla) was able to make a tube of composite materials can return from the atmosphere and land (after a period of learning) without apparent problems and be able to reuse it in a short term. How did they do it?
Lars Blackmore2 graduate of MIT wrote algorithms (Machine Learning) that could learn according to variables related to its environmental environment and make an unmanned vehicle can define its route, reaching its final destination integrated (obviously he works at SpaceX now). Or like Waze that also relies on Machine Learning to define the best route, perhaps without considering many external variables (such as those we have in CDMX). A curious case is what happened in New Jersey, where for a period of time the best route to get to New York was passing through a small town called Leonia, and the mayor decided to close the streets to avoid the increase in traffic. Will the algorithm succeed in learning from man's unpredictable decisions?
Did you see what you achieved in Boston Dynamics3 by applying AI and ML? They designed and programmed a humanoid robot named ATLAS and its puppy SPOTMINI4, which are capable of running, jumping, avoiding obstacles, carrying weights, if they fall they get up and also work as a team. Obviously all funded by the Department of Defense of the United States of America (DARPA5).
Today these technologies are available, maybe not yet for everyone, but it is no longer necessary to have a "Master Degree" in mathematics or be an experienced developer. We have access to many APIs to create applications, services that support neural networks and algorithms of Artificial Intelligence, Machine Learning, or Deep Learning, from Google Cloud Platform to Amazon ML, or such as Microsoft Azure ML, IBM Spectrum, Intel Nervana, among others.
It is important to understand what the differences are since with all these terms it is easy to get confused, we can say that ML and DL are offshoots of AI.
"Machine Learning" means what a computer can learn from what a human does, for example repeating analyses, identifying patterns and applying logical solutions, learning from the result and remembering it.
And with the term "Deep Learning", as a big difference, it is understood that you can learn and write your own code to apply improvements and get the result. Surprising examples were obtained by Facebook where a set of BOTS6 algorithms reached a level where they developed a language that man could not understand (engineers ended up turning off the machines); or what Google7 managed to do, synthesizing the first digital voice without being able to distinguish it from the human one, with ML and DL learning algorithm.
Prestigious universities such as Harvard, license their Deep Learning platform with the private sector to develop new materials for display and lighting with OLED technology.
With "Deep Learning" multinational companies in the technological field are defining our future, solving problems of today's world and making computers act and think like humans; for example, imitating the functioning of our brain, they can analyze a photo or a video and be able to generate a descriptive text of that image, not only of who or what it is, in what place or setting, and all with an accuracy of 93.9%.
And in the digital signage industry? All this is already being applied, and the Mexican Association of Digital Signage (DSMX) is very aware of its evolution and measuring impacts, regulating and promoting the correct application. Audience measurement is based on ML algorithms and Neural Networks, which learn from experience and become more accurate over time.
In Slovenia8, the user experience in a department store is shaped so that ML algorithms are responsible for defining patterns and, with the correct signage, communicating effectively and in real time. All this will become predictive by applying DL, controlling the manufacture and distribution of products, anticipating the needs of consumers, efficiently, without the intervention of man. And as if it were a story of Black Mirror, the machines will be able to create new products, to become creative and one day not too far away, to do without the human being?
*Andrea Mereghetti is CTO of the Mexican company Kolo DS. Text edited by Boris Dallafontana, MKT Manager of Kolo DS.
References:
1 https://blogthinkbig.com/las-6-leyes-de-la-robotica-de-la-union-europea
2 http://www.larsblackmore.com/stochasticcontrol.htm
3 https://youtu.be/fRj34o4hN4I
4 https://www.youtube.com/watch?v=rnTyzwpz4JI
5 https://es.wikipedia.org/wiki/Defense_Advanced_Research_Projects_Agency
6 https://www.forbes.com/sites/tonybradley/2017/07/31/facebook-ai-creates-its-own-language-in-creepy-preview-of-our-potential-future/ - 221b61d9292c
7 https://www.theverge.com/2016/9/9/12860866/google-deepmind-wavenet-ai-text-to-speech-synthesis
8 https://www.researchgate.net/publication/290365887_Modelling_In-Store_Consumer_Behaviour_Using_Machine_Learning_and_Digital_Signage_Audience_Measurement_Data
