In 2029, according to the forecast of one of the top futurists of the world Ray Kurzweil, machine intelligence will surpass human intelligence. We already see modest and courageous prerequisites today. AI changes the world of gadgets, medicine, transport. And even the work of the state. Liga.net and TECHIIA continue the story about 8 technological revolutions in Ukraine during the period of independence. This time we are talking about artificial intelligence. What does it do, and how does it encroach on natural intelligence?
"Come on, artificial intelligence is science fiction." This could be said under the influence of Terminator, who will always be back, or some robot girl from the Ex-Machina movie. But in fact, AI has long been at your fingertips. On a smartphone.
For example, say hello to Siri or Google assistant. Of course, despite all the efforts of developers, the mind of these voice assistants is still far from human. But they recognize our speech easily and call, write, look on the Internet, advise a cafe on our request. Sometimes even tell jokes. And they do all this with their synthetic voice.
The second aspect of AI has already become so common that almost nobody notices it. Thanks to it, smartphone cameras can beautify our selfies and take good photos, even in low light. Two years ago, only the flagship models were "boasting" about it. Recently, the possibilities of artificial intelligence have been "slipping" into the middle and cheap segments. Automatic post-processing of photos is already a must-have for everyone. In the Instagram age, there is no way around.
Yes, AI isn't thinking cyborgs yet. But it is already quite important and applied cases. From automatic subtitles on YouTube to cancer detection, it's better than an experienced radiologist does. From computers that win in Go with world champions to autopilots and smart chatbots.
What do we call artificial intelligence?
Artificial intelligence is a wide-ranging concept. Like in big data, it includes a lot of things. The main direction of this field is the development of technological solutions that work on the principle of human intelligence or close to it. That is, learning, improving, increasingly mastering the common colloquial language (because several leading theories of development say that it is speech and work made a human out of monkey). Nowadays, AI flagships are neural networks, machine learning, data science.
Andrey Yavorsky, VP Strategy and Technology at GlobalLogic:
"It is important to understand that artificial intelligence and all that it means is a certain brain. In fact, these are statistics based on the analysis of large amounts of data, with the help of which AI can help to optimize processes in medicine, automotive industry, finance, retail, and many other areas."
Volodymyr Bandura, CEO of the Innolytics Group research company
"From time to time, there are publications that talk about the death of AI, the new AI-winter, and so on. We work with primary objective data and vice versa — we see significant growth."
From the World International Property Organization Technology Trends 2019, Artificial Intelligence. The grey line is the number of research articles. The black line is the number of patents. These are indicators showing that fundamental research and commercialization of solutions are quite intensive.
According to Volodymyr Bandura, the global trend in almost all sectors of the economy is their digitization. The first step is to collect a huge amount of data, using, already today, billions of sensors, cameras, microsatellites. The next step is processing and decision-making. Accurate operational knowledge of billions of real big and small facts will change the economy significantly.
The further we go, the more data humankind accumulates, and the more powerful processing methods it invents, the closer we get to the real AI. For example, according to one of the world's top futurists Ray Kurzweil, at the end of the next decade, we will lose the crown of the king of nature.
Volodymyr Bandura, CEO of the Innolytics Group research company
"In 2029, Ray predicts that machines will reach and surpass people's intelligence. Not everyone agrees with this forecast. Many equally authoritative experts, such as Sir Roger Penrose, Kai-Fu Lee, propose different and significant arguments against it."
However, we already see some hints, sometimes very obvious, in entirely different spheres of life. What is there now, and what can we expect in the future?
To help a doctor or better than a doctor
Medicine is the top theme for all technologies. Artificial intelligence in some areas already works not only faster but better than humans. Sometimes even notices non-obvious connections. For example, according to Volodymyr Bandura, analyzing the image of the retina AI with 97% accuracy can determine the sex of the patient and even predict the risk of heart disease with high accuracy.
According to Andrey Yavorsky from GlobalLogic, innovations in diagnostics are expected first of all in the next few years. Using smart device data and graphical image analysis (e.g., X-rays), the AI can offer a predictive treatment model and predict the likelihood of a particular disease.
"Medicine will become more and more customizable: treatment will be based on all the information about the person, not just his symptoms at the time of diagnosis," says Yavorsky.
One of the most famous services in this field is Deepmind Health, owned by Google. It analyzes information about the patient's symptoms and generates a list of recommendations to guide the doctor in choosing a course of treatment.
Another service, IBM Watson Health, can diagnose a patient on its own.
"There is a known case where the Watson system managed to diagnose a rare form of leukemia in a patient, while the doctor had previously misdiagnosed him or her. In just 10 minutes, the system analyzed more than 20 million scientific articles on cancer, compared them with the genetic changes of the patient, and, as a result, revealed the correct diagnosis," says Andrey Yavorsky.
But that's just the part that hasn't been widely used yet. The other front is medical devices. GlobalLogic believes that in the next 3-5 years, their autonomy will increase significantly.
The prerequisites are already visible — sales of smartwatches and fitness trackers are growing by tens of percent a year. Gartner analysts expect that only in 2019, 225 million units will be sold, and in 2022 — 453 million. They already track the pulse, duration and quality of sleep. And along with other devices such as smart scales and new generation headphones, they also record other bodily indicators. In the future, it can also supplement our medical records.
"The collected statistics will be actively used for preventive care of patients. After all, it is precisely to prevent and give recommendations in advance - one of the main tasks of medicine of the future," explains Andrey Yavorsky.
Even regular gadgets help doctors. According to Volodymyr Bandura, an American neurologist and AI expert Vivienne Ming presented at the SU Global Summit one of her projects, in which attacks of manic depression are predicted without special measurements, but only by... movement of the patient's phone.
"Based on these movements, the algorithms developed by Vivienne's team learned to predict an attack in a month. This is a very useful result, because these patients have a high suicide rate during attacks, and it is important that at this point they are with someone," says CEO of Innolytics Group.
The third medical front is analytical and theoretical. AI has also found a place here.
Michael Gillam's HealthLab team has programmed standard operations with data used by scientists. The project uses data from hospital databases to automatically generate valuable new knowledge and articles in minutes rather than weeks. Within hours — if you add time for people to check and comment. As Volodymyr Bandura explains, the articles become "alive" and can be updated almost in real-time, with the updating of hospital databases and the addition of new profile cases.
According to Andrey Yavorsky, pharmacology is the slowest moving forward in medicine. For the development of this area, expensive DNA analyses are needed. But in the future, these data will be used to create individual medicines, which will bring as much benefit as possible and as few side effects as possible.
Yuriy Lazebnikov, Managing Partner of TECHIIA Holding:
"Globally, AI is the highest level of automation available to modern society. The consequences are an increase in the efficiency of goods production and the quality of services on the one hand and a decrease in the number of jobs on the other. The demand for specialists with low-level skills will decrease: junior lawyers, cashiers, loaders — they will be replaced by robotic systems. It is necessary to change the approaches to education to give access to rapid retraining. Al will change the paradigm of life in the future."
AI in the car will be an indispensable thing in our lives. The predecessor of artificial intelligence in a car is smart navigation. A system that recognizes objects, signs, markings on the road, and tells where to turn and when.
In combination with the already existing autopilot "without hands" (the driver can temporarily give control to the computer, watching the road), which in recent years has appeared in expensive models — Tesla, Volvo (Pilot Assist), Audi (Traffic Jam Assist), this system can already be called pre-AI.
The application of pre-AI in cars these days looks modest. This is rather a virtual assistant.
"For example, a virtual assistant who studies your habits and is ready to inform you about your plans, routes, and restaurants with your favorite cuisine nearby. Or to call your wife at a particular time at the intersection from which you call her every day on the way home from work. These functions are available in MBUX — an information entertainment system installed in the new Mercedes Benz," says Vitalii Karmazinskyi, Managing Director of Luxoft Ukraine.
As for the trendy concept of the driverless car, it will be developed on a large scale only with 5G. Because, at least, as vendors say, only the communication networks of the fifth generation will be able to provide the artificial intelligence system with the lightning speed of reaction and full control of the road situation.
But, despite the relatively low distribution of 5G networks in the world, driverless cars can already be easily found on the roads of the United States. There are different manufacturers of unmanned vehicles that are now testing them. But, probably, the daughter company Alphabet (Google) Waymo has advanced further than all of them. Within a year, it has been providing an autonomous taxi service in Phoenix (Arizona) now for all people. All the user needs to do is to install an application on a mobile phone.
Robot taxi service is no longer the future, but reality.
There are several categories of tasks for artificial intelligence in the energy industry.
The key issue in energy is balancing production and consumption in real-time on a large territory. With the development of green energy, this problem becomes more acute. After all, if before generation was carried out by strictly defined power plants, now every household can join them conditionally. And it is not so easy to predict the impact of the latter. But it is necessary. Otherwise, the power system will sink into chaos.
"For example, in predicting consumption, existing data are used to train AI algorithms, helping energy suppliers to understand the demand better. Artificial intelligence also allows you to monitor and optimize the efficiency of energy consumption," says Vitalii Karmazinskyi, Managing Director of Luxoft Ukraine.
According to him, more and more often, AI is used to calculate utility bills savings and recommendations for "smart home."
And in the oil and gas industry, artificial intelligence is planned to be used for mineral exploration. ExxonMobil is working with MIT in the AI lab on an ocean exploration robot. Shell is also working on a similar project.
Another area is cybersecurity.
"Attacks on power plants can have great consequences both for the economy and for the defense capabilities of entire nations. Here, artificial intelligence is designed to identify cyber threats and respond to them," says Vitalii Karmazinskyi. Although, of course, this is not just about energy.
The current authorities pay great attention to the digitization of public services. The main task of the new Ministry of Digital Transformation is to put paper permit documents online. And make electronic driving licenses, electronic citizenship, and possibly even the next elections online.
But, of course, all this has little to do with artificial intelligence. After all, public online services are primarily about convenience. The point is the same. It's up to a person to decide whether or not to issue a permit. A person will be chosen in electronic elections, and a person will be at the head of the corner.
But what if you fantasize and imagine the introduction of AI in the management of the country.
First, let's answer the question, how is the management going now? We have a system with a single decision-making center, which has plenary powers — legislative, executive, and judicial power. All members of society agree with this, and they elect those in power. The center collects feedback from the society, turns it into operating norms and initiatives, and then supervises their observance and execution, equality.
Such a system has been used by society for centuries to accumulate efforts, allocate resources, ensure the nation's health and education, insure old age, and, most importantly, to move forward in competition with others of a similar nature.
But the flow of information has accelerated many times over the past few decades. And now the centralized system is starting to fail. Society is beginning to generate more and more complex and frequent requests, and the system, the core of which is the same members of society, i.e., people, is busting from the concentration of incoming energy. Society overthrows the power, "hires" a new one, introduces experienced expats. But in a year it collapses again. After all, they are all humans.
Of course, it is impossible to replace the Verkhovna Rada, the government, and the president with a neural network in one day. After all, it can lead to irreparable consequences. But it is already possible to create such AI and gradually train it in real-life situations.
The authorities continue to make decisions independently, while the machine will process the same input without implementing the results of their analysis. In this way, it will train itself in various situations. In the end, they're all patterns. And sooner or later, you can get a great alternative to the human control system.
It can be assumed that in the future, there will be some service providers who will sell customizable management systems to companies. We need to automate the judiciary — here you are. You want a complete set — anything for your money.
You can read about other commercial cases related to big data and machine training in
Opinions differ on where the development of AI is taking us. Ilon Mask shows solidarity with science fiction writers — artificial intelligence has the potential to kill us. Businessman Jack Ma and futurist Ray Kurzweil are optimists. In their opinion, the development of AI opens new horizons for humanity, rather than crumbling them according to the Matrix scenario.
It's worth thinking about now. But so far, the machine intelligence, although already painting and even writing articles for magazines, is yet limited in its creative abilities.
Well, what's next?
"Try to communicate even with advanced "robots" at exhibitions or online, and you'll realize that before the machines pass the Turing test, they need to jump over the abyss. Will they be able to do that in the next nine years?" asks Volodymyr Bandura.
The question focuses on several problems at once: the amount and power of the computing hardware, the correct algorithms. And, in general, the ability to describe the mind with algorithms. There are fierce disputes on both points.
Yuriy Lazebnikov, Managing partner of TECHIIA Holding
"It is not right to consider the impact of Al on individual industries, because globalization involves the erasure of boundaries not only between countries but also between enterprises. In general, thanks to Al, the role of the human factor, random mistakes and corruption will decrease. There will be a redistribution of points of influence: those who create algorithms will become influential rather than those who have interacted with people before".
As Sergey Bondarenko, the founder of the Semantrum company, said in an interview with Liga.Tech, the world is now engaged in modeling System #1 — human intuition. That is our ability to compare patterns. For this purpose, neural networks train with a massive amount of data. However, we are already working on modeling human logic. And it will already be a real artificial intelligence. Bondarenko argued that it's worth waiting for in the next 10-20 years:
"It will take approximately that long for the digital models of System #2 — that is, our logic — to exceed the 'human' capabilities. This is inevitable, so we need to prepare ourselves and understand how to live in it".