Machine learning and AI: future vs. conceptual mistakes
Any remaining doubts about the importance of machine learning and artificial intelligence should have been cleared up by Google CEO Sundar Pichai in an article written a couple of weeks ago comparing their impact on humanity with that of fire or electricity. Pichai is not a man given to hyperbole: on the contrary, he has a reputation for pragmatism. He is not only very clever, but occupies a position that provides an overview of just where the world is and where it is going.
He highlights a curious paradox: even if we recognize the enormous economic and social importance of AI and machine learning, the gap between ambition and execution in companies in this field is very wide: a survey carried out by MIT shows that although 85% of executives say that artificial intelligence will give their company new business opportunities and a competitive advantage, only one in five outfits have incorporated any kind of artificial intelligence technology as part of how they operate. Only one in 20 has incorporated it fully, and less than 39% of all companies have some kind of AI machine learning strategy. The largest companies, with more than 100,000 employees, are the most likely to have developed a strategy that mentions artificial intelligence, but only half actually have one.
Why this seeming relaxed attitude toward a technology that is going to have an enormous influence on the future? Basically, ignorance. The range of possibilities is so high as to be beyond the average manager’s understanding, making it difficult to set up minimally realistic initiatives in businesses. Instead, we have apocalyptic visions of robots taking over the world: faced with articles that talk about robots being appointed to the board of directors, most middle managers are not inclined to see the advantages and improvements to high-level decision making, but the danger of losing their job sometime down the road.
What’s preventing us from taking a more realistic approach to implementing artificial intelligence? Twitter, for example, has just launched technology able to crop users’ images so as to keep the interesting bit instead of always cutting according to a fixed standard that often makes no sense until seen in its entirety. Will this change the world? No, it’s simply another quick win: the company had a huge archive of images along with accompanying texts that could be used to understand which part of them should be highlighted. Training a neural network to understand which part of a photograph is the most interesting and, therefore, should be kept after editing is a small improvement to its service, but allows the company to see the possibilities of this type of technology. Another algorithm, in Canada, examines profiles on social networks to help prevent suicides, the second cause of death in the country for people aged between 10 and 19 years.
Thinking about the consequences of technological development raises questions about what type of society we are heading toward as machines are able to carry out more tasks: whether this will increase inequality and if we’ll need measures to correct that. We need to think about how people be replaced by algorithms and robots and whether that is a good thing, whether there is grounds for optimism and what responses and policies different countries around the world are implementing, depending on a variety of circumstances.
While these key questions are undoubtedly interesting, it’s clear the world lacks managers and companies capable of understanding AI and machine learning sufficiently to propose realistic projects that generate tangible benefits and that will help them to think about the topic further, to further explore, to develop the area. The time for imagining dysfunctional horror stories with robots and humans fighting for jobs should come to an end, and be substituted with realism and science. Companies that fail to invest in machine learning and artificial intelligence will miss out on opportunities to be more competitive, to increase their turnover, to differentiate themselves from their competitors and to attract and retain hard-to-find talent.
While some are spending their time pondering the future of humanity, others are focusing on the benefits of AI and machine learning, thinking about how to develop tangible projects in an increasingly important field. Beginnings are never easy and projects have a long way to go in terms of defining objectives, data collection and transformation and process engineering, so that by the time you see the importance of this area, you will be way behind the pack, having wasted precious time to develop the possibilities of one of the most important technologies in the history of mankind.
(En español, aquí)
