The global AI market until 2024 will show an annual growth of 17.5% and exceed $500 billion in revenue, as
Doctor diagnostician
Replace humans with robots in healthcaredifficult, but the diagnostician can still transfer part of his tasks to the machine. When making a diagnosis, the doctor relies on his experience and knowledge, but there is always a risk that he will miss something from attention simply due to the human factor. It is easier for AI to manipulate large amounts of data and work with details, which reduces the percentage of errors in analysis.
For example, a startup Zeba can provide assistance.Medical Vision: a service for radiologists and uses AI to study X-ray images, describe and analyze abnormalities. The service has its own database of several million images, high processing speed and accuracy. First of all, it allows you to reduce the burden on the doctor and speed up the process of image analysis.
The larger the database and examples thatlearned by the machine, the lower the risks that it will miss a case. And while AI may not be completely accurate, it is less likely to be wrong than humans. In this case, the final decision will still remain with the living doctor.
Railway driver
Long-distance trains without a driver are unlikelywill appear in the near future, but AI is already beginning to be implemented in the subway. Unlike a living person who drives a railway transport, a computer is immune to errors associated with carelessness and fatigue. This increases the safety of passengers and the safety of goods. AI already controls the train in the Dubai metro, and passengers can be in the "driver's cabin" admiring the view of the city. In New Delhi, they implemented the idea for one of the branches, the Copenhagen metro has long been working due to an automated control system, and in Prague they intend to launch trains without a driver in 2027.
Smart systems are also being developed in Russiatrain management: in particular, Russian Railways is testing 10 locomotives with AI, which are designed to reduce the number of train accidents associated with the human factor. The technology is based on neural networks and computer vision. In the Moscow metro, trains without drivers were expected back in 2017 (Swallows were being developed for the MCC), but this required amending legislation. In 2020, they began to talk about testing elements of automatic train management on the Nekrasovskaya line, but the results are still unknown.
Banking analyst
The key requirements for such a specialist are:deep knowledge in the field of economics and finance, competence in making various types of decisions: from participation in bank investment projects to the selection of loan products for individuals. The work is mainly carried out according to a template, so it can be trusted to the computer.
Most large Russian banks alreadya number of tasks are robotized - at Gazprombank, AI is engaged in card issuance and retail lending, at Rosbank, it processes documents for a client dossier, and at Home Credit Bank prepares personal product proposals. But the most active use of AI is at Sber: all retail credit decisions are made by a computer, 95% of them are generated automatically, without a live specialist.
Shop assistant
The work of a sales assistant is quite routine:he provides clients with information about the company's products and services, helps in choosing and answers emerging questions about the quality, characteristics of the product, and even more subjective: "Is it suitable for me or not?" Similar actions are performed by AI algorithms that take into account individual needs and offer replacement of goods that are out of stock.
The idea was implemented by many clothing stores andperfumery and cosmetic brands. For example, on the Sephora network, Color IQ helps you choose concealer and foundation, and Lip IQ helps you choose lipstick. Back in 2015, Uniqlo had UMood racks showing a few products to the customer. From them you need to choose the ones you like so that the AI and neurotransmitters determine what to recommend to the client. Even direct contact with the device is not required - it is guided by the person's mood. Equally interesting is the idea implemented at the Neiman Marcus department store: the Snap app. Find. Shop. (from the startup Slyce), based on a photograph of an item uploaded by a client, searches for similar or the same products in the store's catalog.
But AI can choose not only clothes, shoes orlipstick - in the online store Instamart (today - "Sbermarket"), which delivers products from hypermarkets, consultants have also replaced machine algorithms. They offer replacements for missing items, analyze customer preferences for previous receipts, and help the order collector speed up the process of filling the cart.
Cashier in a store
The cashier's actions are the same - he punches the goodsfrom the shopping cart, cancels unsuitable items, issues a receipt, accepts payment. AI algorithms can easily do the same thing. First, the fact that a machine can replace a living person was demonstrated by self-service cash registers that appeared in Russia in 2012, and abroad even in the 90s. But today the story has gone much further and AI is replacing cashiers entirely in stores without salespeople.
Amazon was the first to promote the idea with its Amazon Go -smart algorithms see what the customer put in the basket (and what they returned to the shelf), with which they eventually left the store, and deduct the amount of the check from the Amazon account or linked bank card. You don't even need to scan at the checkout. This is not only savings on the payroll (payroll), but also an increase in traffic in the store - 90% of Americans, with a long wait for their turn, will leave the basket and leave. In Russia, the idea was implemented by Sberbank with Azbuka Vkusa (although this is not a full-fledged store so far, but only a department), as well as X5-Group.
Trainer
In a pandemic, the fitness industry finally sawthe demand for online technologies and made sure that it is possible to conduct classes remotely: with the correct setting of the equipment, the coach can work remotely, controlling the actions of the wards. At the same time, it turned out that a computer is capable of replacing a live instructor in many issues: explaining the technique of exercises, monitoring their implementation and the correctness of actions, giving recommendations on the pace, reminding them of breathing, and counting the number of approaches. This is achieved through the use of computer vision technology, which collects information from the camera and transfers it to the machine for analysis.
The idea has already been implemented by several startups - inIn particular, Aaptiv: the company has released a smartphone application with a “smart” trainer that creates personalized workouts and gives lifestyle recommendations. The more often a person uses the service, the more individual his activities become. And Zenia Yoga has created the first yoga app with a virtual AI assistant. It monitors body function through motion analysis of 16 joints, helping the trainer spot mistakes during online lessons or acting as a coach during self-study.
Recruiter
HR specialist searches and selectssuitable employees for vacant positions (actually sorts according to specified criteria). And in the case when the supply exceeds the demand, the recruiter can be guided by personal likes or dislikes in the selection process, and the human factor begins to influence the selection quality. A machine capable of performing the same operations as HR is devoid of the emotional aspect, which means it is impartial and evaluates candidates more objectively.
Pen tests in this direction have been carried out for a long time:In 2016, FirstJob developed the Miyu bot, which offered suitable vacancies to job seekers on the site, connected potential workers and employers. And HeadHunter decided to move its search engine to machine learning in order to make the computer recommendation service more accurate. But everyone is overtaken by the neural network Facebook, which is able to analyze the profiles of specialists registered on the social network, compare candidates, and filter out inappropriate options. It is possible that later it will offer potential employees to companies.
Technologies are also developing in Russia:Vera, a robot recruiter (a startup called Stafory), learned how to select a resume and call employees for an initial interview. In nine hours, he can work out 1.5 thousand candidates (in an interview). And at Alfa-Bank, some of the functions of an HR specialist have been replaced by a computer: now the AI is answering already established employees to standard questions - from "how to order a pass" to "what to do with documents." For this, the AutoFAQ platform was developed based on a trained neural network. The company intends to automate 30% of requests from employees in order to unload HR specialists.
Call center operator
This person receives calls from clients,talks about the company's product, provides technical support or advice. Depending on the size of the organization, the call center can have 200 operators or more than 2,000 (this is the number of Tele2). But even this state is not always enough. Oracle research shows that one in two customers today expects the company to be available 24/7. It is more costly for a large organization to provide this with the help of a staff of living employees than with the help of AI. Therefore, the solution is robots that answer the call and help solve at least typical issues - for example, find out the status of an order. In difficult cases, they already redirect the call to a live operator. But the idea is still at the development stage: robots are trying to collect debts in Russian banks instead of living collectors (nothing is known about the results of the experiment), and in other industries, only 10% of operators are replaced by AI. Although more than half of call center specialists believe that machine algorithms will be able to replace them at least partially.
Chatbots perform similar functions, replacinglive person in chat support. In Russia, they are gaining momentum: in 2017, only 16% of companies implemented them, as shown by the Sherlock.im study, and in 2019 they were already in 60% (Accenture study). It was assumed that in 2020, chatbots will already be used in 80% of companies. It is not known how much the forecasts came true, but at the end of the year, experts and market players indicated that the demand for such AI solutions (classic and voice assistants) had really grown in the pandemic, and in addition to the traditional banking sector and retail, educational institutions, medical, e-commerce were involved. government agencies.
Musician and DJ
It is hard to imagine that a machine could performcreative challenges, however, this future may not be far off. At least in the music industry, smart algorithms are already starting to compete with or come to the rescue of living professionals. In 2017, Yandex created a neural network capable of composing music (and later writing a play, but together with a person). To do this, the algorithms had to analyze huge amounts of data (about 600 hours of music), identify patterns and rules in the musical works of famous composers and understand what affects a person's positive perception of a melody.
Some startups are already trying to developsolutions in this direction: for example, the Chinese Jukedeck creates AI-generated melodies. In July 2019, Jukedeck bought ByteDance - with a high likelihood for the development of TikTok. Similar ideas are being promoted by Amper Music, which was acquired by the Asian giant Tencent. And the Mubert company has collected an extensive database of samples and sounds and taught AI to collect unique musical compositions from them.
The machine has not yet done all the work for composers.will do, but it is already capable of creating background tracks for public spaces, videos and other things where, due to copyright, you can’t just use existing ones - you need to pay the label. In addition, neural networks can generate music for games, as is done by the startup Melodrive and the group 65daysofstatic (who wrote the soundtrack for the game No Man's Sky).
Programmer
In the most digital of industries, the adoption of AI asAn assistant to a person is expected in many specialties, but first of all this will be relevant for programmers. They develop algorithms and write program code to implement a specific task: typical actions, but requiring the processing of large amounts of data. And the sheer volume of program code can be enormous. These tasks can be facilitated by AI, which will be involved at the stage of analyzing and testing the code to find errors and options for fixing them.
In the future, this will speed up development processes.It is possible that the computer will eventually take on the task of writing code, if it accumulates enough suitable examples "in memory". According to experts, low-skilled programmers may disappear - their responsibilities will be taken over by AI.
Taxi driver
The idea of automated cars is being promotedlong. Tesla with its clever "stuffing" gives hope that the era of robotic taxi drivers will come, but so far the world is moving towards it very slowly. In 2015, in Japan, Robot Taxi Inc, together with the leadership of Kanagawa Prefecture, announced a pilot project for taxi robots (automated vehicles) on city streets - previously experiments were carried out on highways. In 2016, Uber was going to launch a similar project, but still with a living person (engineer) driving - this is required by US law. In 2018, Yandex also presented an unmanned vehicle (the experiment will last until 2022). But until now, the idea has not been widely used anywhere. Too difficult task and too unpredictable AI behavior in difficult situations on the road to give him control of a car with live people inside. So far, one cannot do without an engineer who will take the wheel at a critical moment. There is no question of the complete disappearance of taxi drivers now.
The machine will not completely displace humans, because it is capable of working only according to standard patterns, but its connection with living specialists in various professions will strengthen.
See also:
The first accurate map of the world was created. What's wrong with everyone else?
Infrared radiation from human hands was used for encryption
Bacteria found in Death Valley that have been in evolutionary stagnation for millions of years