The "oil" of the modern economy and the war for personnel
All IT giants like Google or IBM have their own laboratories where scientists, engineers and analysts are working on the monetization of artificial intelligence. In 2017, MTS joined the interest of Western colleagues and also opened a division that develops and implements products based on AI technologies. What happens in the "intellectual" laboratory and how it will change the life of subscribers?
I talked with Arkady Sandler , the head of the AI ​​department of development in the field of AI, who has experience in creating various projects in the field of machine learning, in particular in the field of e-commerce. In an interview, Arkady told why AI is the key technology of modern times, what awaits us in the community of a personalized product and how to pump your startup with the help of MTS.
- Arkady, how do you think the current definition of artificial intelligence sounds now?
')
- Artificial intelligence is a class of systems that automatically performs the functions inherent to man. At the same time, a clear definition of the word “intellect” cannot be given also with respect to a person, therefore only “reasonableness” of possible actions for a machine is emphasized here. AI intersects with many other classes of systems, and an autonomous definition for it is still difficult to choose. Now we share the big data systems and systems that automate human functions. Many systems can be used inside the systems, the only question is in their application in each specific case, moreover, each task can be implemented differently. For example, machine learning as a toolkit is used in various fields and is only a sub-branch of AI.
- How difficult is it to develop and “train” a neural network, what resources are needed for this?
“Nowadays, students of specialized specialties can study neural networks, especially since universities are increasingly connected with market demands, and there are more and more programs for working with data and machine learning.
In the topic of AI, a key factor in measuring the scale of work is the availability of data that has been collected, marked up and can already be used to train a neural network. In an industrial format, only a team of specialists can process a large amount of data.
- Where did HIP come from around AI?
- I would use the metaphor that the data are now the "oil" of the modern economy. Data ownership is a critical factor in global power. Therefore, the giants of Silicon Valley invest in AI billions and open all new units to work with data. In the coming years, major technology companies will fight for personnel who work with data and machine learning, and develop areas of work with AI, for example, in terms of forecasting and automating business processes.
- If there is enough data, then artificial intelligence can learn everything?
- Rather, if there is a sufficiently large amount of data, you can build a system that will master this skill and be able to perform "humanly." It is the quality of the data that is important. But in the concept of AI, there are two ways of applying data - this is general AI and narrow AI, and now the latter prevails on the market. General AI is not yet a very achievable idea of ​​a comprehensive and global AI, which will really be able to do everything and resemble the human mind.
- It has already created a phobia that the AI ​​will soon deprive us of work. What changes can the AI ​​actually bring to the labor market?
- Each industrial revolution, and we are now inside the next of them, leads to a significant restructuring of the labor market. Naturally, the widespread introduction of systems based on AI will lead and is already leading to this restructuring. Some professions die off, but new ones also appear. Now there is hardly a factory where cars would be manually assembled. Humanity does not seem to suffer from this, but, on the contrary, has more and more opportunities to consume and self-realize. AI-based systems relieve a certain number of people from routine work.
People can engage in creative work, go to creative spheres and self-realize, and not spend 8 hours a day on mechanical actions. This is initially an incorrect paradigm - to drive living people, capable of various actions, independence and creativity, into a system of elementary tasks. Therefore, it is worth looking at the changes only positively.
- In what area can you find the actual problem that the AI ​​can solve?
- Usually, when one or another initiative connected with an AI is created, for example, a startup, we think abstractly. Imagine that you have an unlimited army of free labor at your disposal, but each of these people is able to perform only one cognitive function. For example, read the text, and based on the content, press one of a number of buttons. It is necessary to present it - and immediately there will be a task for this army. Good goal finding exercise.But if you are a businessman and understand where the breach - and this is exactly the area that AI can patch up - then you can open a startup.
- Is artificial intelligence capable of learning?
- This question requires about a month of scientific reasoning. AI is able to learn. The ability to self-educate in understanding a person is, to put it mildly, exaggerated. But there are areas where this is possible, if we talk about narrow AI, not general.
- Turning to the recent experiment with the publication of the first book of neuropoetry , I would like to ask - where do the ethical limits of the use of AI lie, if any?
- If the network writes poetry as well as Pushkin, then the ethical question arises not about the network, but with respect to the author of the experiment. Ethics are the prerogative of a person, while the system merely imitates its cognitive functions. In my opinion, in the era of the narrow AI, ethics and AI do not overlap, and the true limit of ethicality lies alongside the “culprit” of the experiment.
- What does the division of AI in the MTS?
- We conduct applied research in the field of artificial intelligence and develop products - for example, smart bots or products in the medical and legal fields. There is also a place for purely scientific cases, we are now actively cooperating with universities and are open to joint projects in the field of education. These will be the points of intersection of scientific and applied interests. In addition, this is a good opportunity for students to immerse themselves in the field of urgent tasks for AI.
- What specific tasks are facing the AI ​​team within the company's requests?
- The modern company is full of ways to use AI systems, and we are exploring new ways. We receive tasks from the customer service department, and we help in servicing subscribers, we also developed a contact center robot and invent ways to simplify the processing of documents for our lawyers. In general, we optimize the work of the divisions where we eliminate the human factor and make the results more effective.
- How will AI change the product directly for the client?
- Having studied the consumer, AI makes the product more personalized. In general, consumption personalization is the future that we are entering through the development of AI. This frees us and allows us to approach everything, including consumption, creatively, and therefore takes us away from industrial goods in the direction of copyright. We are waiting for the boom of independent entrepreneurship, in part this is already happening. We like bread from a small bakery more than from a supermarket, clothes of a local designer, craft beer and coffee from an eco-plantation. Personalization will definitely win the bare practicality of industrial products.
- Did you cooperate with such independent entrepreneurs within the framework of the work of the AI ​​subdivision? Shared technical developments?
- We are constantly working with startups as part of the MTS Startup Hub accelerator, which selects the most interesting projects in our opinion to help in development and cooperation. Together we make pilots and, of course, mutually enrich our knowledge. We are particularly interested in projects related to the field of Natural language processing (NLP) - this is one of the basic families of AI technologies, and we have a high level of expertise in this area. But in general, open to any opportunity to cooperate in the field of AI.
- Now you are preparing for the project with the HSE Business Incubator - what awaits the participants of the AI Startup Accelerator , which will start soon?
- We will advise young startups in the field of AI as experts. The most interesting of them, MTS is ready to take in the pilot. We are ready to take on perspective startups, to give them the opportunity to develop so that they in turn can move various industries.
For the opportunity to talk and help in preparing the interview, special thanks to Marina Morozova
I talked with Arkady Sandler , the head of the AI ​​department of development in the field of AI, who has experience in creating various projects in the field of machine learning, in particular in the field of e-commerce. In an interview, Arkady told why AI is the key technology of modern times, what awaits us in the community of a personalized product and how to pump your startup with the help of MTS.
- Arkady, how do you think the current definition of artificial intelligence sounds now?
')
- Artificial intelligence is a class of systems that automatically performs the functions inherent to man. At the same time, a clear definition of the word “intellect” cannot be given also with respect to a person, therefore only “reasonableness” of possible actions for a machine is emphasized here. AI intersects with many other classes of systems, and an autonomous definition for it is still difficult to choose. Now we share the big data systems and systems that automate human functions. Many systems can be used inside the systems, the only question is in their application in each specific case, moreover, each task can be implemented differently. For example, machine learning as a toolkit is used in various fields and is only a sub-branch of AI.
- How difficult is it to develop and “train” a neural network, what resources are needed for this?
“Nowadays, students of specialized specialties can study neural networks, especially since universities are increasingly connected with market demands, and there are more and more programs for working with data and machine learning.
In the topic of AI, a key factor in measuring the scale of work is the availability of data that has been collected, marked up and can already be used to train a neural network. In an industrial format, only a team of specialists can process a large amount of data.
- Where did HIP come from around AI?
- I would use the metaphor that the data are now the "oil" of the modern economy. Data ownership is a critical factor in global power. Therefore, the giants of Silicon Valley invest in AI billions and open all new units to work with data. In the coming years, major technology companies will fight for personnel who work with data and machine learning, and develop areas of work with AI, for example, in terms of forecasting and automating business processes.
- If there is enough data, then artificial intelligence can learn everything?
- Rather, if there is a sufficiently large amount of data, you can build a system that will master this skill and be able to perform "humanly." It is the quality of the data that is important. But in the concept of AI, there are two ways of applying data - this is general AI and narrow AI, and now the latter prevails on the market. General AI is not yet a very achievable idea of ​​a comprehensive and global AI, which will really be able to do everything and resemble the human mind.
- It has already created a phobia that the AI ​​will soon deprive us of work. What changes can the AI ​​actually bring to the labor market?
- Each industrial revolution, and we are now inside the next of them, leads to a significant restructuring of the labor market. Naturally, the widespread introduction of systems based on AI will lead and is already leading to this restructuring. Some professions die off, but new ones also appear. Now there is hardly a factory where cars would be manually assembled. Humanity does not seem to suffer from this, but, on the contrary, has more and more opportunities to consume and self-realize. AI-based systems relieve a certain number of people from routine work.
People can engage in creative work, go to creative spheres and self-realize, and not spend 8 hours a day on mechanical actions. This is initially an incorrect paradigm - to drive living people, capable of various actions, independence and creativity, into a system of elementary tasks. Therefore, it is worth looking at the changes only positively.
- In what area can you find the actual problem that the AI ​​can solve?
- Usually, when one or another initiative connected with an AI is created, for example, a startup, we think abstractly. Imagine that you have an unlimited army of free labor at your disposal, but each of these people is able to perform only one cognitive function. For example, read the text, and based on the content, press one of a number of buttons. It is necessary to present it - and immediately there will be a task for this army. Good goal finding exercise.
- Is artificial intelligence capable of learning?
- This question requires about a month of scientific reasoning. AI is able to learn. The ability to self-educate in understanding a person is, to put it mildly, exaggerated. But there are areas where this is possible, if we talk about narrow AI, not general.
- Turning to the recent experiment with the publication of the first book of neuropoetry , I would like to ask - where do the ethical limits of the use of AI lie, if any?
- If the network writes poetry as well as Pushkin, then the ethical question arises not about the network, but with respect to the author of the experiment. Ethics are the prerogative of a person, while the system merely imitates its cognitive functions. In my opinion, in the era of the narrow AI, ethics and AI do not overlap, and the true limit of ethicality lies alongside the “culprit” of the experiment.
- What does the division of AI in the MTS?
- We conduct applied research in the field of artificial intelligence and develop products - for example, smart bots or products in the medical and legal fields. There is also a place for purely scientific cases, we are now actively cooperating with universities and are open to joint projects in the field of education. These will be the points of intersection of scientific and applied interests. In addition, this is a good opportunity for students to immerse themselves in the field of urgent tasks for AI.
- What specific tasks are facing the AI ​​team within the company's requests?
- The modern company is full of ways to use AI systems, and we are exploring new ways. We receive tasks from the customer service department, and we help in servicing subscribers, we also developed a contact center robot and invent ways to simplify the processing of documents for our lawyers. In general, we optimize the work of the divisions where we eliminate the human factor and make the results more effective.
- How will AI change the product directly for the client?
- Having studied the consumer, AI makes the product more personalized. In general, consumption personalization is the future that we are entering through the development of AI. This frees us and allows us to approach everything, including consumption, creatively, and therefore takes us away from industrial goods in the direction of copyright. We are waiting for the boom of independent entrepreneurship, in part this is already happening. We like bread from a small bakery more than from a supermarket, clothes of a local designer, craft beer and coffee from an eco-plantation. Personalization will definitely win the bare practicality of industrial products.
- Did you cooperate with such independent entrepreneurs within the framework of the work of the AI ​​subdivision? Shared technical developments?
- We are constantly working with startups as part of the MTS Startup Hub accelerator, which selects the most interesting projects in our opinion to help in development and cooperation. Together we make pilots and, of course, mutually enrich our knowledge. We are particularly interested in projects related to the field of Natural language processing (NLP) - this is one of the basic families of AI technologies, and we have a high level of expertise in this area. But in general, open to any opportunity to cooperate in the field of AI.
- Now you are preparing for the project with the HSE Business Incubator - what awaits the participants of the AI Startup Accelerator , which will start soon?
- We will advise young startups in the field of AI as experts. The most interesting of them, MTS is ready to take in the pilot. We are ready to take on perspective startups, to give them the opportunity to develop so that they in turn can move various industries.
For the opportunity to talk and help in preparing the interview, special thanks to Marina Morozova
Source: https://habr.com/ru/post/447910/
All Articles