TheAI(Artificial Intelligence) consists in the ability of a machie to adapt to a given context by formulating autonomous solution strategies: to do this, it acts by translating the above context into algorithms– formal procedures that solve problems through a finite number of steps – more or less complex, depending on the evolutionary level that distinguishes it.
The AI field is not necessarily related to robotics, as most robots currently in circulation do not need extraordinarily advanced software and hardware. The conversation changes from the moment we start talking about service robots, designed to interact with humans by communicating, learning, and collaborating.
In recent times, thanks to a technology developed in an almost exponential way, what was previously relegated to the world of science fiction is becoming more and more realistic, attracting ethical debates. A curious fact – disturbing to some – that recently happened within FAIR(Facebook AI Research) concerns the anomaly detected during a “negotiation” between two AI prototypes. The experiment was based on testing the ability of the two machines to complete economic negotiations using human language. Everything went smoothly until the pair of computers began to communicate – and make decisions – by moving away from the anthropic idiom for which they had been programmed, and using a language unknown to the programmers themselves. The test, at that point, was aborted and the anomaly corrected. In any case, this event, documented and disclosed by the experimenters, has raised the concern of many. The idea that two machines can “plot” without the knowledge of man, in fact, alarms not only laymen and fans of sci-fi comics, but also people like Bill Gates, Ellon Musk, Steve Wozniak and even Stephen Hawking! The latter, for example, has alerted the scientific community by highlighting the huge gap between the biological/human evolutionary times and the virtual/digital ones of a machine that, according to the astrophysicist, if equipped with intelligence and started freely to the development could make giant steps in very short time, completely escaping the control and causing incalculable damages. To get an idea of the extent of such damage, think of robots equipped with the ability to self-repair, self-program and self-reproduce: devices capable of reasoning and acting at speeds tens, hundreds, thousands of times faster than ours “Sentient” devices, able to foresee human actions and reactions, exploiting our every weak point in order to subdue us, exploit us and in the end, obviously, exterminate us. Actually it is not difficult to imagine: literature and cinema, for decades, propose similar scenarios in all sauces.
Even Googlehas recently spoken out against the use of artificial intelligence in “decision-making processes”, but its reasons are much more technical. In the tech giant’s view, choices based on automated pattern detection within large sets are not neutral, but resort to patterns based on the very biases that determined the composition of the information sets.
Despite these and other concerns, global investment in AI has been experiencing strong growth in recent years, and for Accenture(a U.S. multinational management and strategy consulting, technology services and outsourcing firm) by 2035 they will lead to an annual doubling of development in 12 advanced economies and improve productivity by up to 40 percent.
It was Googleitself, in 2014, that absorbed startup DeepMindfor $400 million, one of the most significant acquisitions in the history of the industry. Generous investments also from Spotify, which merged several specialized companies in order to refine content suggestion and the targeting of ADV messages on its platform. The same goes for Amazon with its 15 machine learning centers (one of which in Turin), and eBaywith its recent acquisition of StartUp specialized in the sector… all the biggest market giants, in short, are moving in this direction.
Artificial intelligence is also advancing in the world of financial trading: according to the forecasts of the consulting firm Tractica, revenues obtained thanks to AI-based software will reach 60 billion dollars in 2025, most of which will be achieved through financial transactions. The operation of artificial intelligence, we have previously seen, is based on increasingly sophisticated algorithms and in 2016, according to a study carried out by Aite Group – an independent research and consulting firm with a focus on business, technology and regulations and the impact they can have on the economic services market – 66% of the world’s nominal equity value was traded precisely by algorithms; processes now capable of autonomously building real-time predictive systems and choosing winning strategies in the face of changing contexts. But that’s not all: iSentium(a US startup founded in 2008) has developed AI algorithms even capable of consulting social networks such as Facebookand Twitterand, through logical, grammatical and semantic analysis of texts, assigning values and scores to the notions obtained in order to establish trading strategies.
Another sphere of use of AI in the financial field is the one dedicated to report writing: a synthesis operation carried out on exorbitant amounts of numbers and information (company balance sheets, macroeconomic data, etc.), usually carried out as a “gavetta” by young financial analysts. Today, with the use of logarithms, such reports can be written by a computer at the speed of thousands of pages per minute.
But what is possible now with artificial intelligence is still nothing compared to the changes it will bring about over the next 35 years, according to the study Harvard and Yale – on advancing the sophistication levels of intelligent machines – discussed at the ThINK Digital, organized in Milan by GroupM with the support of The European House Ambrosetti, which was attended by some of the leading experts in the field. By 2024, AI systems will be translating foreign languages better than humans, and by 2027 they will be driving buses, trucks and cars with mastery superior to any driver on the road. In 2049 they will be able to write compelling best-sellers and in 2053 they will perform the first surgeries with complete autonomy.
Looking into the near future, 2018 would seem to be the breakthrough year for artificial intelligence. According to a survey by IDC(a global group specializing in market research, consulting and event organization in the IT and TLC fields), 28% of companies have already adopted AI-like solutions, while 41% intend to proceed on this path within two years. Opinions regarding potential job losses are divergent; US Treasury Secretary Steve Mnuchin, for example, assures that AI will not pose a threat to American employment over the next few decades, but on the contrary will offer new job opportunities born from human-machine collaboration. The international management assistance company McKinsey, on the other hand, after analyzing two thousand activities in 820 types of work, has come to the conclusion that in 60% of them at least 30% of the functions can be automated, but that in any case, currently the professions that can be totally automated do not exceed 5%. Moreover, the growth of the productivity given from the aid of the artificial intelligence in the analyzed industrialized Countries, between which Italy, is in a position to climbing from the 0.8 to the 1.4% every twelve months.
The only certainty is that the race towards AI and automation has begun and, unless it is interrupted by some unpredictable cataclysm, it seems to be unstoppable.
At this point, it is absolutely useless and detrimental to be scared out of one’s mind: instead, it will be necessary to prepare and educate oneself, modifying one’s cultural and working structure in order to take advantage of the revolutionary development possibilities that will be offered to us in the near future.
Marcello Argenti