Artificial intelligence is no longer a discipline relegated to research laboratories. Applications are widespread and there are several examples where different AI technologies are implemented. But before illustrating some areas, we want to remember the two elements that are, on the one hand, making it possible to spread AI applications in business, and on the other, make them available in real-time where they are needed:
To make the use of large-scale AI applications in business possible, to "democratize" access by making them usable even for small and medium-sized businesses and not only to large corporations, is the spread of cloud computing. The use of AI technologies and applications from the most widespread public clouds, in which the powerful processing mass required to run AI applications takes place, also opens the door to those who do not have great investment opportunities.
The second fundamental element is edge computing: in applications where it is necessary to have an immediate response to problems that can be solved with artificial intelligence applications, these must be performed where the data is collected, with peripheral processing (and for this reason they are such important technological developments as the ReRAMs.
In this context, the potential for application is wide, thanks to everything that can be done with the analysis of big data and the clinical history of patients applying machine learning: from improvements in the diagnostic phase to the possibility of administering personalized care based on to the genetic makeup of the individual.
Great are also the promises of artificial intelligence for the prediction and prevention of diseases or epidemics on a large scale. There are so many e-health applications. These range from remote monitoring of health conditions thanks to wearable devices, to routine testing without medical intervention, up to the calculation of the probability that a patient is suffering from an illness, not to mention the support that can provide studies in the genetic and genome fields.
Hospital structures can benefit, but also the health system of entire countries, thanks to the reduction of hospitalization costs. A market survey conducted by Accenture in 2018 estimates that US healthcare by 2026 could save 150 billion dollars thanks to artificial intelligence applications both in areas closely related to the medical activity (from general medicine to surgery, to drug administration) both in terms of information security in hospitals.
Tractica, a company specializing in market research in the AI and robotics sectors, estimated in 2017 that global revenue from medical imaging technologies should reach $ 1,600 million by 2025, while global revenue from virtual assistance apps could, even by the same date, exceed $ 1,200 million.
One of the best-known applications of artificial intelligence in the automotive world is that of autonomous cars: according to the financial giant BlackRock in 2025, 98% of vehicles will be connected and in 2035 75% will be self-driving.
The application of AI in the autonomous guide does not fail to raise ethical problems. MIT researchers published in October 2018 in Nature the results of the survey The Moral Machine experiment which involved 2 million people in 233 countries to understand what people think about the choices that a self-driving car should make in an emergency. If the car has to choose whether to crash (in danger of killing the driver) in order not to invest a child, what should it do? And if instead of the child there is an elder? What do you choose between investing in a group of people or a single person? Or if on one side there is a homeless and on the other a well-dressed lady? If on some answers (better to save people than animals, larger groups compared to smaller groups) the opinion was fairly shared on others cultural differences emerged: in Latin America, for example, it is preferred to save young people than the elderly,
But it will not be necessary to wait for the cars that drive on their own to see the AI implemented on the cars : one example is smart video cameras equipped with facial recognition systems, already installed on trucks and commercial vehicles: they detect the driver's status by monitoring fatigue, distractions, states of poor lucidity (allowing, in addition to avoiding accidents, to lower insurance costs for commercial fleets) and some car manufacturers are developing solutions, less expensive, to be implemented on cars.
Finance and stock market
According to research conducted by Accenture in April 2018 (which saw the involvement of 100 CEOs and top managers and 1,300 employees worldwide, in the banking sector), 76% of CXO banking institutions believe that the adoption of AI technologies by 2022 will be a critical success factor in this market.
With what impact on financial institutions and what are the advantages for customers? These are the questions to which he tried to give answers to the huge study of August 2018 made by the World Economic Forum in collaboration with Deloitte, The New Physics of Financial Services - Understanding how artificial intelligence is transforming the financial ecosystem, from which we take some cues.
The study finds that there are very many specific financial services that can gain important benefits from the adoption of artificial intelligence technologies. As shown in Figure 5, these services have been grouped into 6 areas: Deposits and loans, Insurance, Payments, Investment management, Capital markets (all activities related to the management of financial instruments issued by bank customers such as bonds, shares and derivatives) and Market infrastructure (brokerage of brokerage services for example derivatives management). Each service then positions itself on a maturity scale that starts from the use of the AI to "do things better" (A) to arrive at use that allows "to do things in a radically different way" (B).
The AI is a new battlefield on which to play customer loyalty. The AI enables new ways to differentiate its offer to customers based on extremely punctual customization of the services that can be offered in real-time when a market opportunity or a customer need arises.
Professional firms, from lawyers to accountants to notaries, are considered a sector that will increasingly be impacted by artificial intelligence technologies not only to automate routine activities, but also for activities of medium complexity, replacing human work which, at best, it can be retrained for higher-value activities but at worst it will be excluded from the sector until the disappearance of some professional figures present in these studies.
We briefly summarize some of the activities that could benefit from the adoption of AI technologies:
Document review and legal research: AI software can improve the efficiency of document analysis for legal use, cataloging them as relevant to a particular case or requiring human intervention for further investigation; due diligence support: searching for information on behalf of their clients with "due diligence" is one of the most demanding activities of legal assistance studies; it is a job that requires confirmation of facts and figures and a thorough evaluation of decisions on previous cases to provide effective support to its customers. Artificial intelligence tools can help these professionals conduct their due diligence more efficiently; contract review and management. Labor lawyers perform very demanding activities to review employment contracts to identify any risks for their clients; the contracts are reviewed, analyzed point by point to advise their customers if they need to sign them or how to renegotiate them. With the help of machine learning software, it is possible to draw up the "best" contract possible.
Let's get to some examples. Although published a couple of years ago (and in this field two years are almost a geological era) the Can Robots be Lawyers studio , created by two economists from MIT and the North Carolina School of Law, shows that 13% of the work of a lawyer can be automated: the model illustrated in the study is based on machine learning algorithms that allow you to automate the different activities of a law firm (from document management to law analysis, to writing the clauses of a contract, two industry, up to the preparation of debates, etc.) by referring to two types of instructions, one based on data (which is "grounded" by the algorithm) and the other is deductive.