No matter how much the media tries to persuade us, robotization cannot lead to mass unemployment. Automation takes time and considerable expense, and artificial intelligence itself has much to learn from people. Moreover, history shows that technologies, that caused certain professions to disappear, has also helped create new jobs. We discussed this last time. Nevertheless, some of the tasks performed by people are easy to automate. As a consequence, certain jobs no longer require as many employees as they currently employ. In this article, we will look at specific examples of new technologies that appear in various industries and occupations, and whether they truly can replace people.

Who is at risk?

Researchers of Russian career portal Superjob^[1] believe that starting from 2017, the demand for entry-level accountants will decrease, and by 2020 the number of these specialists may drop by 3 times. Banks will start to get rid of employees involved in paperwork. Automation will gradually take on the work of contact centre employees.

From 2018, the demand for qualified employees of industrial enterprises will begin to decline. It is also forecasted that there will be a reduction in the demand for foreign language teachers (at the same time there’s a lack of qualified business translators, which is going to grow even more) and mining workers. In addition, starting from 2018 the demand for low-skilled workers will begin shrinking by 5% annually.

Then who benefits from that?

According to ManpowerGroup^[2], an increased demand for specialists in information technology (IT), human resources (HR), and customer relations is expected in the future.

In the short term (for 2017), analysts of Superjob predict an increased demand for specialists of the following professions:

IT: mobile developers, web developers, information safety and cybersecurity specialists, Big Data analysts with additional experience in other subject areas.

Mining: exploration and development specialists, technology advancement specialists.

Manufacturing: mechanical engineers, including oil & gas engineers, forestry specialists, aviation engineers, food processing engineers.

Sales: sales agents of services and high-tech products.

Banking: managers, including anti-crisis professionals; mortgage and troubled debt specialists .

Legal advice: international law and tax specialists.

Human Resources: HR directors, analysts with technical background, corporate training specialists.

How will robotics affect various industries?

There’s general agreement that any monotonous work is doomed to automation. Today, robots are already able to perform actions which, as was recently believed, require human evaluation and perception. But robots are still far from achieving a total takeover of the labor market: in some areas, machines lose significantly to people. Robots still cannot work without a certain pre-defined methodology; they are not good at the assessment of people’s emotional state or understanding the context of some phrases. A robot can’t respond properly in various situations: for example, it won’t offer help or fetch a glass of water for an upset person.^[3]

Nevertheless, the fourth industrial revolution affected most, if not all, spheres of human activity. In some cases technologies simply facilitate the work process, in other instances, they easily complete this process without human interference. Let’s  describe the most curious examples below.

Industries:

• ACCOUNTING
• AGRICULTURE
• CALL CENTERS
• CONSTRUCTION
• GOVERNMENT
• HEALTHCARE
• HOSPITALITY
• INSURANCE
• LEGAL ADVICE
• LOGISTICS/TRANSPORT
• MANUFACTURING
• MEDIA
• PUBLIC CATERING
• RECRUITMENT
• RETAIL

Accounting

Last year, Deputy Minister of Finance of Russia Tatiana Nesterenko stated^[4] that the profession of accountant should gradually disappear from the market and be replaced by technology.

But even if Russia already got used to electronic reporting, the technology of electronic document exchange with business partners, which has existed for some time, is not that popular now (just ask yourself how many of your suppliers in Russia have already switched to it).

Agriculture

American Case IH^[5] presented an autonomous tractor project. However, in fact it is not completely self-driving: a remote operator is supervising it and coordinates its pathways with the help of desktop computer or tablet.

Similar developments are being conducted in Russia^[6]. By the end of 2017, Cognitive Technologies should present a prototype of an automated control system for agricultural vehicles.

Call Centers

Contact centre employees could be replaced by chat bots. Pypestream^[7] claims that their bots can replace about 80% of support staff. The system analyzes the customer’s request and, if it is a well-known and frequently-asked question, responds automatically. Uncommon and complex client issues are redirected to human employees.

Sberbank’s Iron Lady does a good job with the routine work of calling credit card debtors. This is not just a pre-recorded voice: the system recognizes natural human speech and responds to it accordingly (Iron Lady also uses machine learning, so the system vocabulary will be expanded while communicating with customers). And if a conversation does not work out, it’s possible to switch from a robot to a human phone operator. In the future, developers plan to make Iron Lady “softer”: it will learn to understand a person’s emotions and to express them itself.

Construction

The use of 3D printers in construction is no longer a novelty. The technology of building with 3D printing already allows not just printing separate elements or blocks, but entire homes, which results in reduced demand for workers (e.g. McKinsey estimates that 64% of work in Russia’s construction industry could be automated^[3]). As an example, let’s take the experience of Apis Cor^[8]. The mobile 3D printer of the Russian start-up managed to print self-bearing walls, partitions and a building envelope with an area of 38 square meters in 24 hours (by the way, the cost of construction was only RUB 600,000).

Another robot, SAM100^[9], was created to work together with the mason (and not for replacing people with robots), and according to the developer’s promises can increase productivity by 3-5 times.

Government

The use of artificial intelligence in the public sector is discussed in the report Reform thinktank^[10]. According to the researchers, chat bots will be able to replace 90% of UK officials. Experts at McKinsey Global Institute share more modest figures: according to their estimations, activities of UK administrative and governmental institutions can be automated by only 37%^[3].

For example, the local council of Anfield (London) decided to use a virtual assistant Amelia to answer questions from the residents^[11]. It understands the context of a conversation, applies logic, learns, resolves problems, and even tries to understand the emotions of the one who communicates with it. So far it has a secondary role — no one is going to lay off people — and the virtual assistant is used only to help with routine tasks.

Healthcare

Some of doctors’ work can be performed by computers. Surgeons are already using automated systems, for example, Telegraph reports that carrying out operations with the help of the da Vinci robot has helped to reduce patients’ blood loss and improve the their overall post-surgery state.^[12]

Artificial intelligence has also proved its efficiency in diagnostics: IBM Watson is showing great results in lung cancer detection by analyzing the results of MRI.^[13]

In the video below you can learn about PillPick. This pharmacy automation system helps hospitals with the process of sorting, storing, and dispensing medication. However, the system is not likely to replace the human employees completely, but to aid them.

We hope that the free time (for example, for Russia it is 41%) made available to medical staff in connection with the use of new technologies will be spent on improving the quality of treatment and patients care, and will not end in the reduction of employees.

Hospitality

In 2015 Japanese hotel Henn-na Hotel introduced a new “strange” hotel format^[14], where human presence was reduced to a minimum: robots meet guests at reception, take their luggage to the hotel room, place things in the cloakroom area, and serve as a concierge. Human employees just have to watch for security and make beds in rooms.

Given that in 2017 the company opened a second robotic hotel, we can assume that the idea of replacing personnel with robots was successful.

The overall automation potential for Japan’s hospitality is 71%^[3]

Insurance

Big Data and machine learning may deprive insurance agents of work.

Japanese firm Fukoku Mutual Life Insurance has already replaced 34 employees with the IBM Watson Explorer AI system^[15]. The technology is expected to not only to save the company $1.2 million annually, but also to increase productivity by 30%. By the way, the cost of installing the system is about $1.8 million, and annual maintenance will cost $135,000.

The automation potential of Japan’s insurance industry is estimated at 58% (43% for Russia^[3]). Well, things do not look very optimistic for insurance workers.

Legal Advise

Lawyers are well-paid highly qualified specialists, but, as it turns out, they are also predicted to disappear from the labor market.^[16] What does technology offer?

1. The online platform ArbiClaims allows residents of Los Angeles to settle small claims disputes (up to $10,000) faster and easier, and it only costs from $79 to $159 (depending on the value of the claim).The service offers the plaintiff and the respondent a court hearing in the form of an online conference with one of ArbiClaims’ lawyers acting as an arbitrator. If the defendant refuse to attend the plaintiff has to solve the matter in the old way — by taking the dispute to the small-claims court.ArbiClaims helps to settle disputes over loan repayments, past due invoices, car repair, security deposits, and services contracts, and their decision is judicially enforceable.

   If you think about it, the service does not actually deprive lawyers of their work: ArbiClaims says that their arbitrators spend 60-90 minutes to deal with one case, instead of 10-15, which are spent by ordinary judges. (The facts are studied more carefully before a final decision is rendered). At the same time, the service greatly simplifies the life of plaintiffs, respondents, and witnesses: they don’t have to leave office during business hours to go to court, and wait there for their turn. ArbiClaims allows all parties to connect from any place at any time convenient for everyone.Result: clients are happy, and lawyers are employed.

2. JPMorgan Chase & Co implemented software, which does in seconds what the company’s lawyers spent 360,000 hours per year doing^[17]. The program called COIN performs the tedious job of interpreting commercial loan agreements. In addition to high speed, it is also less error-prone.Online service Legal Robot can scan your documents for common legal inaccuracies, possible risks, or simply translate from legalese to English (this is the only supported language so far). The developers themselves say that some of lawyers’ work is quite monotonous and boring and can be easily automated.There are similar services: Lawgeex and Kira. The latter, for example, helps lawyers reduce the time spent for legal Due Diligence, analysis of contracts and lease summaries. Developers assure that the platform can save 20-90% of the time spent for these tasks.In any case, each of these services is more aimed at improving quality and reducing errors than replacing a real person with a machine.

3. The NYT mentions the American ROSS system, which significantly reduces the time to search for relevant information in legal databases^[18]. AI Lawyer’s abilities already include understanding English; formulating hypotheses on the given issues; search for information in databases and generating answers with references and citations on the basis of which the conclusions are drawn^[19].

4. Here is an example from Russia. Sberbank no longer needs 3,000 lawyers^[20]: robots will allow the bank to automate the filing of claims against individuals. Employees who used to do this work will be either retrained or dismissed.

5. According to the research^[21] carried out by Dana Remus, the University of North Carolina, and Frank Levy, Massachusetts Institute of Technology, even if all the new technologies suitable for the legal field were immediately implemented, they would have only saved 13% of lawyers’ time (Researchers from McKinsey Institute estimate that 23% of lawyers’ work can still be automated). In the most likely scenario, over the next five years, automation will annually reduce the required number of working hours for lawyers by 2.5%.

To sum it up: artificial intelligence proved its usefulness when searching in a database for necessary documents required for a specific case, searching for mistakes or for other monotonous work. However we have to agree that such lawyers’ duties as consulting clients, preparing legal grounds, negotiating and representation in court, are currently difficult to automate.

Logistics/Transport

Amazon is one of those companies that actively implement robots in their activities to automate the process of assembling and packing orders. Among the warehouse robots there are “babies” half a meter tall, able to move around the warehouse at a speed of 8 km/h and carry loads of up to 317 kg, and huge robotic manipulators designed to move heavy pallets. At the same time, Amazon does not intend to use robots everywhere: machines are used only where it is truly cost-effective.

In December 2016 an Amazon drone performed first package delivery to a customer^[22]: a streaming media player and bag of popcorn ordered from the online store arrived by air to the buyer 13 minutes after the order was placed. The growth in popularity of this unusual method of delivery is hampered by a number of restrictions: it is possible only if there is a spacious landing area located near the Amazon depot, and the package cannot weigh more than 2.6 kg. So far only two customers can test the delivery by drones, but Amazon expects to increase this number to several dozen. Although it is assumed that air delivery should be available 7 days a week, the drones will be able to deliver only during the day and only under good weather conditions.

By the way, the air couriers work completely autonomously — additional human control is not required.

It is possible that in the long term such a method of delivery will become popular when sending goods to rural or remote areas.

Alphabet holding is interested in creation of a marketplace for drone delivery^[23]: the IT giant is negotiating with Domino’s Pizza, Whole Foods, and several other fast food companies. Incidentally, there was an attempt to start cooperation with Starbucks, but due to disagreements the relationship did not work out.

Russian restaurant chain «Додо Пицца» (“Dodo Pizza”) conducted an experimental drone delivery as early as 2014^[24]. However, a number of questions remain unsolved such as who should retrieve pizza from the device and hand it over to the client. In the experiment the parcel was received by a company employee, making this delivery method not completely automated.

Mercedes, General Motors, Uber, OTTO, Google, Ford, Tesla, KAMAZ, and dozens of other companies are developing self-driving cars. Their plan is to allow buses, cars, trucks and even ships move from point A to point B completely autonomously and without any human involvement.

So far, the systems of self-driving cars have a number of flaws: in certain cases the driver has to take control himself. Nevertheless, researchers at Altimeter assure that a fully autonomous car (5th level of autonomy according to SAE classification, while what we see today is at most 4^th generation autonomy^[25]) could appear as early as 2021^[26]. Forbes, referring to a report of Boston Consulting Group^[27], predicts that by 2025 the percentage of fully autonomous cars will be no more than 0.5%, and 9.8% — by 2035.

Let’s not forget that whether this forecast will come true or not depends on a number of factors:

1. Development of technology itself. Artificial intelligence must learn to make independent decisions not only in a number of familiar situations, but also in unforeseen weather and road conditions. Cars need more accurate sensors and extensive databases with all sorts of situations on the roads; vehicles must be able to exchange information with each other and infrastructure.

2. Cost of technology and final product. It is expected that, in addition to improving road safety, self-driving vehicles should further reduce the cost of passenger transportation as well as the service fees of transport and logistics companies at least because of savings in personnel. But this will be possible only with the appearance of a fully autonomous vehicle, which means the cost of sensors (and other technical equipment) and software should not significantly increase the price of such a car.There are already cases of significant cost reduction for some components: a few years ago the cost of the LIDAR sensor was $80,000, and today this sensor costs $8,000; Velodyne plans to further reduce the price to $50.However, one shouldn’t forget that even with the advent of completely autonomous vehicles at a price of a regular car, companies will need to resolve the issue with scrappage and recycling of old model cars — something that can’t be done in a short term.

3. Ethical issues. How will AI make a decision in case of emergency: should it save the passengers inside, passengers of another car it may crash into, or a family at a bus stop that those cars may hit? Difficult question, isn’t it?Who will be responsible for damage: software developer, technical engineer, manufacturing company, or a person in a driver’s seat who could take control?Is it possible to blame a person if he does not have driving experience?

4. Privacy. Using self-driving vehicles, you allow the system to accumulate a huge amount of data about your routes and whereabouts. In the most inoffensive form, a computer can use this information to remind you that today you forgot to buy your favorite coffee and offer to make a regular purchase. But can you see what a large window of opportunity it creates for hackers and other criminals?

5. State regulation and public opinion. The first can be a consequence of the second. Issues of ethics and security, fear of losing jobs — all of this can affect government regulation of traffic by self-driving cars.

Any of these points can postpone both commercial and civil use of autonomous vehicles. Their initial use will likely to be possible in very limited cases.

Nevertheless, the potential automation of transport and logistics industries in Russia is estimated at 75% and in the long term, technologies possess a real threat of dismissal of 4.3 million employees.^[3]

Manufacturing

Automation is an integral part of modern manufacturing. It is broadly used in automotive industry, food industry, and other fields. Robots perform monotonous, hard, and often dangerous work: welding, painting, assembling, moving, etc. And factories are not going to stop there. Due to further automation, manufacturing companies plan to improve output and product quality, and, of course, reduce costs. Employees are at disadvantage here — new technologies do not require a large staff for maintenance.

We already mentioned several similar instances in a previous article, and here is another example from China. Before introducing automated assembly lines Changying Precision Technology^[28] factory (mobile phones manufacturer) used to employ 650 employees. Now only 60 employees work there — all the rest is done by robots. As a result, productivity has increased by 250%, and defects has dropped by 80%.

According to McKinsey, 64% of work process in China’s manufacturing industry can be automated (54% for Russia).^[3]1:0 for robots: in this industry their use may continue to grow, and human labor required — to decline.

Mass Media

According to Forbes, journalists can be replaced by tools that are able to rewrite other people’s content, with sports and financial journalists being most at risk.

Such news agencies as Associated Press and Bloomberg are known to use Wordsmith^[29]. This product, created by Automated Insights, allows to generate a huge amount of text from a single story structure and data set.

Public Catering

73% for USA, 72% for Germany and 67% for Russia — a few examples on estimations of the automation potential for public catering sector. Now let’s look at the technologies that can replace human workers.

Tired of waiting in line to get your favourite coffee? Perhaps you should try ordering one from a robot barista of Cafe X in San Francisco. It seems that everything is great, but don’t you think that this is an excessive complication of what works so well? Yes, undoubtedly, this invention is impressive, but what’s wrong with good old coffee vending machines?

A few years ago, Momentum Machines introduced a prototype robot that could cook a whole burger^[30]. If you believe the news, they will soon open their own automated fast food restaurant. We’ll wait and see.

BratWurst Bot is able to prepare grilled sausages. One can’t say that the robot is graceful, but it manages to do its work. The only question is, does the manufacturer allow changing the functionality of the robot without significant expense? After all, in this case the functionality is very limited.

The robotic kitchen from Moley Robotics is more impressive: two humanoid hands skillfully perform the actions of a cook. Commercial sales are scheduled for 2018 (in 2015 the device was estimated to cost $75,000), but for now you can watch the promo video:

A simpler example — Spyce Kitchen^[31]. Automation allows mixing and preparing a number of ingredients, while special sensors monitor the desired temperature and quality of food.

There’s pure harmony in the kitchen of Zume Pizza^[32]: robots and people work as a perfect team. One of the robots, called Martha, is responsible for spreading the sauce on the dow, and the other, Bruno, sends the prepared pizza covered with toppings by its human co-worker, into the oven.

We found several cases of how restaurants use robots to replace waiters, so that they attract and greet visitors, take and bring orders. But the example of Guangshou’s restaurant chain proves that you should not rush to replace employees with androids^[33]: the owner had to close two restaurants and leave only one robotic waiter in the third one. It turned out that robots, for example, were not able to bring soup or other liquids without splashing it, and often broke down. Their functionality was strictly limited, so no matter how visitors asked, the robot could not pour a glass of water for them. The only thing that the unusual employees were able to do well was to attract customers.

Recruitment \ Headhunting

Forbes says that recruitment of candidates, interviewing, and all related document flow can be automated^[34].

Let us think how it may look like. Imagine a software that can search for candidates with certain criteria on different job portals and websites (LinkedIn, Indeed, etc) and assess (in %) how a candidate meets the requirements of your vacancy. As a result, HR manager needs only a few seconds to get a scored shortlist of suitable candidates.

This platform also enables video interviewing with a system that analyzes candidate’s emotions and truthfulness of his answers. Just imagine that during an interview the system can identify if a candidate lies when answering your questions.

When all the interviews with candidates are done, HR Manager can put his own scores for each candidate in the system. That means that machine learning algorithms will improve the system scoring and it will be more accurate next time.

You can say that sounds like science fiction and that will never happen. But let’s check what tools exist now.

Applicant tracking systems: that is a software that recruitment agencies use now. It helps to organize and plan the process: search for resumes on different job portals, import resumes from various sources, post vacancies on different job sites, schedule interviews, prepare forms for assessment/evaluation, etc.

Or have a look at amazinghiring.com. This is a search engine that can find IT specialists with required skills among dozens of social networks and websites.

GoRecruit platform can analyse social media profiles and put a score for a candidate for a chosen vacancy.

Facial Expressions recognition software is also not a fiction. Computer algorithms can now identify microexpressions that are hard to notice for a human eye^[35]. This means that it is going to be much more harder for job seekers to hide the truth in the future interviews.

Do you still think that such comprehensive software that combines all mentioned technologies is impossible? We suppose that this is just a matter of time. When it appears it can dramatically change the entire recruitment process that exists now.

Retail

Amazon Go^[36] is an experimental store where you won’t find any salespeople, checkout, or lines (none at all!).

When you enter the store, a scanner reads the QR code from the buyer’s smartphone, and the technologies similar to those used in self-driving cars start to do their job: computer vision, sensor fusion, and deep learning (in other words, all your actions inside the store closely watched and analyzed). As you go the system notes which products were taken from the shelves, and which ones were put back. When you exit the store, the mobile application automatically charges money for purchases from Amazon account and sends you the receipt.

This one-of-a-kind store is located in Seattle, USA and operates in beta format. Currently only Amazon employees can shop there. The company promises to make the entrance public at the beginning of this year.

Watch a short promo clip of the futuristic supermarket below.

However, according to the latest news, not everything is going so smoothly: the system may lag when there’s a big a crowd of shoppers inside.

SIMBE ROBOTICS offers a robot that performs the routine work of retail workers.^[37] Tally informs store employees of out-of-stock and low goods; misplaced goods; and incorrect price tags. Tally is not a substitute for human employees, but it helps to save more time and use it for better customer service.

Experts estimate that 52% of work in Russia’s retail can be taken by robots.

No matter how intimidating these robots can be, most of them will not take our jobs. Some of them will provide efficient support for monotonous everyday tasks. They will increase productivity, reduce human errors and save more time for employees to spend for important things (e.g. training).

At the same time, some bots can cause a real revolution in certain industries. But one shouldn’t forget about the number of factors influencing the speed of automation integration process: technical capabilities, development and maintenance costs, labor costs, economic environment, state regulation, and public opinion; all of these have a direct impact on labor market as well. For now history has shown that replacing manual labor with machines and disappearance of some occupations did actually create a great amount of brand new jobs. That is why we sincerely hope that  Fourth Industrial Revolution won’t be an exception.

Related articles

PART 1: Robotization and Automation: Impact on the Labor Market