Alexander Vityaz (Corezoid): Chatbots, as a separate kind of software, do not exist
Alexander Vityaz is a landmark figure in the IT community. For 17 years, he was head of CEB (Centre for Electronic Business) at PrivatBank, where advanced banking products were developed: from Privat24 (in its four guises - SMS, mobile, web and chat) and LiqPay payment system to P2P transfers and QR-code payment. We are all used to SMS authentication, for example, which is now used by all banks and messengers, but it was PrivatBank that first implemented it. Since 2017, Vityaz has been running its own company, Corezoid Inc. whose core business is a proprietary development of the same name. Describing the Corezoid system in two words is quite difficult. At first it was a kind of lego, which now would fit under the term zero coding, then the creators themselves began to call their brainchild an operating system. Finally they came to a concept "orchestrator", which also requires explanation to an untrained person (we talked about it with Alexander Vityaz).
Corezoid's monetization models are SaaS and OnPremise, the modern and understandable for businesses - a subscription or purchase of a software license. Corezoid Inc's customers include large banks, retail chains and multinational corporations. Although Corezoid can also be used by a sole proprietorship, for example to create a chatbot. Actually we were going to talk specifically about chatbots, but the outlook and vision of Alexander Vityaz is so wide (e.g. he is convinced that everything in the world can be described through actor graphs, i.e. to build and count connections between all elements in the system, about which he wrote the monograph "Codex of Actors") that the conversation predictably covered much wider topics - from changes in programming market with emergence of AI and voice interfaces in banking applications to low autonomy of augmented reality glasses and the rise of energy communism.
gg: How can companies benefit from chatbots?
Alexander Vityaz: Let me start with a joker - chatbots as a separate type of software do not exist. What in everyday life is called a chatbot is an application that has a "front end" in chat. The essential part of any application is the "back end" and the "middle end", the logic and architecture are there. If you have the former and the latter, you have everything to connect any 'front end' (or several at once). For example, if you're best at calculating sin, then which fronts enter x and output the result sin(x) is not crucial, it's purely client preference. Ideally, the client should be able to seamlessly travel between different fronts.
A chatbot is a "front end" with a "middleware" inside and a "back end". It's what the customer interacts with. If you design what's under the hood properly, the number of fronts can be unlimited. We focus on algorithms. And in fact, if you do it ill-conceived and start designing for a chatbot, the architecture is unlikely to be suitable for the application. You'll have to redesign everything when you want or are ripe for the application. And if you look at the fact that chatbots are one of the "fronts" possible for an algorithm, then you'll get everything in "back" and "middle" right. And the bots, the app, the website - whatever - will exist peacefully.
There is no question about the usefulness of chatbots - any communication is useful and any way of communicating with the customer is good, the more communication surface of the company the better. The question is about communication management, measurement and decision making. Of course, if you have 5 clients it is not rational to write an application or even a bot.
Tasks can be solved with chatbots because the algorithm is not in the chatbot, but in the backend. What's more - using a bot as an application front-end is cheaper and faster than developing a native application
gg: Judging by the previous name of your company, you specialize in middleware, aren't you?
Alexander Vityaz: We help build bridges between "backend", "frontend" and external software. Our earning model is license sales. The client signs up or buys a full license. At the same time, we provide services for the implementation of all this, help get you started, provide training, and so on. Our speciality is the digital core (middle, back and everything in between and the front end). We've managed to make such a platform that it gradually absorbs the entire enterprise software infrastructure. We didn't intend it that way, but that's the evolutionary way, if you build a circulatory system, sooner or later you have to build a heart and a brain.
Illustration: Corezoid like orchestrator
gg: Is your company is synonymous with integrator?
Alexander Vityaz: There is a more universal term - service orchestrator. It is used by our clients, it is an industry term. Integrator is too narrow a term. I just connect it and then what? An orchestrator is both about connection and management. It's not enough to connect all the cables, you have to manage it all further. Among the names we had were the director and the conductor, but it didn't take. At first it was a conveyor belt, which turned into a service orchestrator, then an operating system, later we started building a digital core, after which we moved to the level of creating an actor processor (Actor Engine). Now we created the ability to combine actors into a graph of actors, and Corezoid became an orchestrator. The meaning has definitely been preserved, but the level of understanding of what Corezoid builds has increased qualitatively.
Illustration: Corezoid like ERP (Enterprise Resource Planning)
gg: How do you evaluate the chatbot market in Ukraine?
Alexander Vityaz: There is no point in measuring it, but in the limit it equals the number of live businesses. I would phrase the question differently: how many companies need a communication with the client? All of them. And all employees, too. That's the answer to the market size question.
gg: And what is the share of chatbots in your company?
Alexander Vityaz: Let's change the word "chatbot" to "front". There is always a place for the "front" in our company's work, because the "backend" and "middleware" have to communicate with the consumer somehow and show something to someone. But what it is and what proportion of one or another kind of frontend is situational. We deliver a digital core to 100% of our clients, someone connects only a chat, someone an application, someone all at once.
gg: If IT in this picture of the world is always connected with communications, it turns out that you are globally involved in the process of automating everything that can be automated?
Alexander Vityaz: Not exactly. Because not everything can be automated and not everything needs to be automated yet. In this context, a communicator or messenger could also be called an asphalt paver. After all, when we build a road, people have an opportunity to go somewhere, for example, to visit each other and communicate. So, we can regard the road as a messenger, too, to some extent. Whether it is automated or not is a second question. Somewhere it is possible, and it does not harm communication. Somewhere it is difficult.
My questions are hardly answered by the same GPT. But if I address my questions to some professor, he will answer me better than GPT today. Better so far.
That said, GPT's call centre operator level questions are already very decently answered
It's important to create channels of communication, easily and quickly. But how to communicate there? How much automation there is in them is a question of management. For example, we have already decided for ourselves today that the first level of helpdesk will be answered by GPT. And the second level - if a client is not satisfied with the answers of the first level (for example, his rating is bad in NPS), we pass it to a person. This is quite normal.
gg: What prevails in platforms for chatbots in our country - Telegram or Viber?
Alexander Vityaz: It all depends on a segment, somewhere Viber is #1, somewhere Telegram. It is right to connect all channels at once and give the customer a choice.
gg: What kind of companies come to you?
Oleksandr Vityaz: All types. From private entrepreneurs to large corporations. The biggest client has a capitalization of 200 billion. Subscription to corezoid.com starts at $30 per month.
gg: What are the main skills a typical entrepreneur needs to use Corezoid?
Alexander Vityaz: What is a private entrepreneur? It is not a profession. If it is an engineer in the form of a private entrepreneur, it is immediately clear what he needs to do. If a hairdresser is a private entrepreneur, it will probably be difficult for him. You need the skills of an engineer and a trained business analyst. MySQL, a little JavaScript. You can use it without any of this knowledge, but you can't really do anything worthwhile. Need some kind of engineering background in the form of specialized education or practical work experience. The need depends on the problem you are solving.
If you need to make a simple robot, of course, you don't need an engineer for that.
If it's a complex project with a million algorithms, you will need a highly skilled engineer.
gg: What are the steps involved in developing a chatbot?
Alexander Vityaz: The company needs to know two things - to have industry expertise and to go to corezoid.com, the rest we will help, we will show, teach, launch and accompany. We provide a turnkey communication orchestrator (not to be confused with the service orchestrator), it allows you to set up all communication channels chat, email, voice, forms, comments, add gpt, and then manage clients, operators, bots, content and event calendars.
gg: What advantages does Corezoid offer for creating chatbots?
Alexander Vityaz: All in one - front-to-back, billing, anti-fraud, AI, CRM, payments, training and outstanding expertise. With Corezoid, chatbots have been created at Sinevo, FUIB, Metro, UZD, Ibis and many others. Of the latest, kai.ai is a very useful and promising psychological help project.
gg: What are the difficulties and risks of creating and launching chatbots for business?
Alexander Vityaz: There is an architectural risk - if you design it wrong, the plasticity of the solution will be low and the price of improvements will be high.
gg: How do you measure a chatbot's success?
Alexander Vityaz: The usual business metrics: number of clients, transaction, dialogue score, NPS (customer loyalty index, Net Promoter Score, a survey methodology by which customers are asked to rate from 1 to 10 the probability that they would recommend the company to their environment - editor's note). With the advent of Large Language Model (LLM), we will introduce indirect evaluation, for example by the tone of the client's speech.
gg: So it turns out that chatbots only make sense if you have a large audience?
Aleksandr Vityaz: Chat-bots are not only about commerce, but also about communication. It's a channel convenient for customers to ask a question. But a chat-bot is still a command line communication format. If you need to upload, for example, a thousand items, perhaps messenger is not the most convenient format for this.
In our world, chatbots do not exclude apps. They complement each other, chatbots are called from the app, apps are called from the chatbot. The right answer is to do everything, they are communicating vessels.
For some kind of experimentation, hypothesis testing - it is clear that chat is the cheapest channel.
You quickly put together an algorithm in the backend, run the bot, check how customers react. And then you put it into an application. As a sandbox for experiments, an MVP, a chatbot is the right solution. Because in this case you don't have to deal with the "front end". And if you're making an app, you have to work on three fronts at once: Android, iOS, people darling and all that stuff. The price of a mistake if you translate an idea into an app right away is probably tens of thousands of dollars and months of time.
gg: If someone decided to launch a chatbot for his company, what would you advise to start with?
Alexander Vityaz: It's very easy to give advice (laughing). If someone comes up with exactly the idea of "launching", that's trouble. Because he'll give a command to make a chat-bot. Then something gets stitched together, and then the real life begins - the refinements, costs and other discoveries. In other words, this chat-bot (application, website - it doesn't matter) will be a constant distraction, and it all becomes a point of expense at once. But to turn it into a revenue point, you have to do some serious work and thinking. A programmer won't do it for a manager.
What is the solution? Think, design. If you don't have experience, you either have to go somewhere, or you have to raise your own specialist. Serverless is the right paradigm. It certainly hasn't evolved to the right form, the way it was conceived. But the concept of cloud hosting, in theory, is correct. You used to maintain the server yourself - buy it, fix it. Breeding those DevOps around. Now everything has moved to the cloud, but there aren't fewer DevOps, by the way. Because the DevOps in clouds have not happened yet. That is, I buy servers there anyway. Yes, I don't keep them at my place, but I still think in terms of servers, OS versions and other processors. This is a big problem.
But someday I won't think in those categories. The industry is moving in that direction, but it hasn't reached the finish line yet. I will work with some kind of database or knowledge. And not worry about questions - what kind of processor is there, what kind of server and how much memory I bought or not.
gg: Do you know anything about using chatbots in education in Ukraine?
Alexander Vityaz: Our clients use chatbots for training their employees and clients. But, again, training is about content and algorithms, not the front end.
In education we have a phenomenal spilno.school client, be sure to read up on where and how they are moving.
gg: Why do you think there is so much talk about chatbots in recent years?
Alexander Vityaz: Because the majority of messenger users have this experience - bots have been around for decades and people are used to them. Because there is a legend about easy entry into IT. There are two myths - one about testers, the other about chatbots. As easy as it is to get in, so easy is it to get out. If you're not a professional, you won't stay in the profession. Some businesses go into messengers because that is where their audience is. Some go because they don't have the budget for a full-fledged app. Apps are expensive. They are designers, iOS developers with huge salaries and so on. But there are times when you have the money and the app, but the audience lives in chat rooms - it's situational. We have clients who have more traffic in chats, and there are clients who have the opposite: chat is auxiliary, while the main traffic is in the app.
Talking about low entry costs is speculation
Is there a low cost of entry in medicine? And it applies to any profession. You get in, then what? If you don't have the ability, you get out in the same way. It is clear that courses that teach programming need to sell themselves somehow. And the message about easy entry - it works. But it's almost a hoax.
gg: Then what do you say about zero coding?
Alexander Vityaz: Zero Coding is now doing ChatGPT. Is it necessary to learn it at school if computer already does the job better?
I'll tell you that we've already fired several people from jouniors, javascripts and designers because GPT does their job better.
gg: Here we have a lot in common with journalism...
Alexander Vityaz: And how is a programme, in terms of language and linguistics, different from a text? Both are languages with different syntax. Text in a programming language is a subset of natural language. That's why (to simplify) GPT is so easy to program, it's easier than mastering a natural language.
Zero Coding exists as a phenomenon, but it doesn't exist as a profession; it is Zero Coding, because everybody can use it.
For example, what is an engineering profession? You have to know mathematics. There is the profession of a translator, there is the profession of a linguist. Which one of them knows the language better? The linguist. And the translator is a better user of the language. Question: if we say that a development engineer is a translator, then we can throw out the word "engineer". And if we say that he is an engineer, then he should be more of a linguist. There is a lot of mathematics in linguistics, actually. It is mathematics, because it's structure and trees and everything else.
To the question "Can you be a developer without knowing maths" the answer is you can't. You can pretend to be, which many people have done with success. But I never liked these expressions - low code and zero code, because the point is lost behind them. With Corezoid you don't have to know any language specifications. But you have to know the maths, you have to have the subject matter expertise to do a complex algorithm. And if you don't have the ability or the knowledge, you won't make anything. That is, you can do a simple process, but not a complex one.
When I was at university, my profession was called applied mathematics, mathematical engineer was written in my diploma. And at that time there was not yet a profession in its pure form as a programmer. But the structure of this profession was that first you learn mathematics and at the same time you learn to program on the computer. But the DevOps profession (development & operations - the methodology of development of complex software, which is constantly being improved in the process of use - editor's note) was in general a neighboring department. Because the task of an applied mathematician is to make a mathematical model, an algorithm for solving the problem. For this, in fact, apart from a head, pen and paper, you don't need anything.
To distinguish between programmers with and without engineering training, the market invented the word coder. Sometimes it has an indecent prefix :-).
Of course there are self-taught guys who have mastered the basics of the profession, but they are more the exception than the rule.
gg: It turns out that a chat-bot developer is a function, but not an existing profession?
Alexander Vityaz: In fact, it exists on job sites, but, in my opinion, it's just a kind of job for a programmer. There is a market. For example, some company wakes up and writes in an ad, "I want to develop a bot. Some people come there. When unqualified customer and contractor meet, what they can do is likely to be "not good".
"Developing" chatbots is easy to become and so there is no prospect of such a "profession". It is an oxymoron like "smartphone user". But there is an engineering profession, and developing chatbots is one type of work that an engineer does. You have to become an algorithmist, for that you have to take at least a basic Computer Science course.
gg: What advice would you give to students - what they should study now to be competitive in the future?
Alexander Vityaz: The advice has not changed for 20 years - study mathematics. Our difference so far between what we call artificial and natural intelligence is in being emergent. In the ability to create and apply some knowledge and generate others. Here in the GPT today, there's not even close to intelligence. But nothing will stop it from emerging. As long as there is a niche - connecting the dots in a graph, creating knowledge.
Translating a JavaScript terms of reference like Google Translate is not a job. The job is to write the terms of reference.
For example, if we used to get 10 responses per week for a javascript job, now, after GPT, we get 200. The difference is 20 times. Of course, there is also a military factor, because some companies are closing, relocating, but this trend has intensified since the advent of GPT. Salaries have not fallen yet. There are more entry-level specialists on the market, but there is less need for them. For an HTML coder I don't think there's going to be a need for it again, most likely.
gg: Does it turn out that programmers will deal more with architecture in the future?
Alexander Vityaz: But that's actually how they came about. The father of today's explosion with large language models, or rather one of the fathers, is Judea Pearl. He invented Bayesian networks. That is, a mathematical apparatus, which is now finally getting a real implementation. But when he came up with it, he was drawing formulas. He wasn't sitting at his keyboard at night, typing. He created a mathematical device that has only just been implemented, because computing power has developed sufficiently.
gg: Have chat-bots reached the peak of their development?
Alexander Vityaz: They haven't, on the contrary, it's just starting. Introduction of GPT will make communication qualitatively better and applications more diverse.
Large Language Model (LLM) + direct knowledge = expert system (ES)
I'll tell you more - without any linguistic models we had bots living in PrivatBank's call-center since 2005. And 70% of customers who talked to them thought they were talking to a person. These bots worked simply by keywords. You take the statistics of the questions, write the answers, the keywords, and everything works fine. If a person needs to find out his balance, he does not need a professor to answer the question. One way or another he will write in the chat "what is my balance", "what balance do I have". Some kind of synonyms. When I saw the stats, I was amazed. The Large Language Model works even better. But even with dumb automation, the Pareto rule was almost respected. Almost 80 per cent of clients didn't understand whether they were being answered by a human or a robot.
gg: Does your helpdesk actually work with ChatGPT and responds without scripts?
Alexander Vityaz: Operators actually have scripts too. It's not a freestyle - whatever he wants, he answers. There is product documentation, manuals which you can leaf through, FAQs, timings and so on. We now have a thing called the Communication Orchestrator. By the way, to understand the evolution - when we first started working with chat rooms and bots, a long time ago, it was called a bot platform. But for two years or more now it's been called our communication orchestrator. And this evolution of the name shows how much our view has changed - from the particular to the general. And so today we have artificial intelligence standing in that orchestrator as a collaborator. And the one who sets up this orchestrator puts it in place of a real employee. And any language model can be plugged in, not necessarily GPT - Google, Facebook or several at the same time. Whichever one is on the market. The client sets up an account, enters the keys and connects the intelligence they need.
gg: Isn't there a risk that the GPT will give some kind of answer that is irrelevant or even harmful?
Alexander Vityaz: Control it, you can check it before launching, test it. Isn't there a risk that the chat operator will write something wrong? There are a lot more costs for a human to manage the whole operation. Yes, the robot has to be monitored, but so do people. So in terms of managing communication, nothing changes.
gg: What new opportunities for chatbots have appeared in the last year?
Alexander Vityaz: Multichannel, BPM, ES. Multichannel - Apple, Google, WhatsApp, Viber entered the messenger market for businesses. This has increased the requirements for proper architecture, quality, usability, design and generally raised the bar - you have to try very, very hard to become a partner. BJM makes it possible to better understand customer issues and create new products in messengers - for example, expert systems. This, in turn, reduces the requirements for bot developers - any advanced PC user can create a simple ES answering questions on documentation in a couple of minutes.
gg: What has changed with ChatGPT? Is it already a game-changer or not yet?
Alexander Vityaz: It creates a new interface. In a broad sense of the word. An enterprise is some kind of function, an algorithm. It changes wheels, delivers food, something else, whatever. It implements some function. And in order for you to use this function, you need some interface. An application, a bot in natural language, not in natural language. Some forms and so on. But our native interface - the language - makes the appearance of a large language model a lot easier.
For example, there are 500 pages in a banking application. Adding each new feature means more pages, people, tasks, documentation, "backends" and so on. Now imagine tomorrow a 501st service is added. And we don't update the application. Because you consume this service through a bot.
These interfaces didn't appear today - Siri, that's it. IVR voice robots in call centres are their forerunners. But they did not have a large language model, which means that Siri could barely understand anything. And also with our accent, it didn't understand anything at all. But that's precisely the challenge of the interface that the big model solved.
No matter how you twist the phrase: "buy milk", "go and get milk", Siri will only understand two phrases out of ten possible. And the large language model will understand all of them.
Now anyone can do them. An expert system of some kind, for example, for your novice colleagues. If you have an instruction, you give it to the large language model, and it can work.
LLM is our new eyes, ears, language and memory. A few years ago I said that it would become possible to communicate with the masters of the past on the basis of their work. (But there are prospects for the mass emergence of proper interlocutors: there will be Chekhovs in the clouds, neural networks built on the basis of works, letters and other preserved texts. We will have direct access to the greatest minds, then you journalists will be able to do any interview and no one will refuse you - interview from 2019).
gg: Does it tell the consumer some quote or the right answer to the question asked?
Alexander Vityaz: It's even more interesting - you give her a seventh-grade geometry textbook, and after that she can solve seventh-grade problems. It is no longer a question of a literal understanding, but of understanding the rules of use. And that's not difficult. It is still not intelligence, but the interface has changed qualitatively, because you no longer have to give hard commands. What is the difference between a natural language and a programming language? The rigidity of syntax. For a computer to understand you as it exists today, you can't do without rigid syntax.
And the average programmer is a translator from open syntax to rigid syntax.
But someone is communicating information through this syntax. That someone is a real engineer. Someone who tells you how and what to do. What is the gap in the classical programming paradigm anyway? The product owner is not an algorithmist, but he or she, for example, changes currency. But the programmer is not a financier and does not know the word futures or the rules of the market. And so they are trying to find a common language between them. And this is a problem. With the large language model this gap will be eliminated, we will all become algorithms and problem setters. Because the computer already understands almost all of our messages.
gg: In the near future should we see some voice functions even in banking applications?
Alexander Vityaz: They've tried it before, but failed miserably. In order to make a normal voice control applications you should have been engaged in what GPT was doing for 15+ years (I mean the company OpenAI). And to have appropriate class of specialists and financing. For Ukrainian company it is unreal to have such a budget and access to such people. And this concerns not only Ukraine, but the majority of countries and companies. Today we see that in the world, at most 10 companies are engaged in it. That is not dozens of them, especially not hundreds. Because it's an expensive pleasure, gpt only for hosting costs hundreds of thousands of dollars per day.
gg: Speaking of new interfaces. You used to make a Privat24 application for Google Glass. What do you think, is there a future for such glasses, for example if Apple will make something that will change the world?
Alexander Vityaz: Google has shut down their glasses project. There is no future for those glasses. But there is a future for augmented reality. But how it will be implemented is an open question. Most smartphones do not have very good battery life today, and glasses are even worse. That is why, when we go somewhere, we immediately look for a charger. It is almost an instinct.
I don't know what it will be, the question is not about glasses per se, but about power sources. As soon as they become miniaturised (or wireless), some unequivocal, if not revolutionary, movement in its direction, will happen. Battery life is not enough. We see the same thing today in smartphones, in electric cars, everywhere. Cars have already reached the point where you can drive (although not yet in the same way as gasoline).
gg: When someone says that they have developed an AI chatbot, what do they really mean?
Alexander Vityaz: Before the advent of FYM, I have no idea, they probably exaggerated (laughing). Often the answers were passed off as AI by keywords. It is realistic to use BCN today for question recognition, ES creation, call centre automation.
gg: Do you have a vision of how AI will affect humanity? Is it optimistic or pessimistic?
Alexander Vityaz: Of course I do, and who doesn't? (laughing) First of all, what is optimistic or pessimistic. I know you have cats. The question is: do you control them or do they control you? And now, instead of cats, put yourself. Judging by the pictures, your cats are happy.
Life is already changing. But not everything can be extrapolated. When a person goes to the first grade you can extrapolate his life to the institute that will happen to him. Well, give or take. Now imagine that you can't extrapolate. We can fantasize. What is our planning horizon today? Who planned six months ago that layoffs of juniors, javascripts and call centre operators would start? Nobody.
Most people don't really work intellectually, they work physically.
So physical work will remain, and intellectual work won't be for everyone. But everyone doesn't need it. The question about salaries and incomes will somehow be solved, because the economy will not go anywhere, it will be transformed along with labor.
The book "Dr. Einstein's Monsters" reports that the model of the universe can be calculated by 100 people in the world, the complexity of the models doubles every 20 months. Proxtrapolate further on your own.
The monthly active audience for GPT in March 2023 was about 100 million.
Double that, adjusting for the smart but principled, you get 200 million. There are 8 billion people in the world. Only 2.5% of people passed the curiosity and intelligence test. Of course, over time, the number of users will grow due to easier access, embedding into familiar interfaces, etc. But in this context it is the primary impulse that matters - only 2.5% wanted to immediately touch the aggregated knowledge of all mankind. This is to the question of how many people work with their heads and how many work physically. Practically, the reverse of the Turing test has occurred - a computer is testing humanity for intelligence.
A robot, on the other hand, doesn't need Coca-Cola, trousers, mascara. And when a proper, real, autonomous artificial intelligence (A.I.) emerges, part of the economy we are used to will start to change.
gg: And the more robots there will be, the more people will be needed to create and maintain those robots?
Alexander Vityaz: Robots and AII are two big differences. Robots are the material physical presence of the A.I., and there is an infinite number of them needed. There will be several A.I.'s at the start - by the number of thermonuclear countries, and in the limit 1. In science it is called a singleton, there can be only one. AII without quantum computers will not appear. A.I. is very much related to energy. First, there is a lot of counting to do, second, the calculators need to be heavily cooled, and third, it all needs to be physically and cryptographically protected.
Robots are the next smartphones and electric cars, whose main brain is in the cloud. As long as they are serviced by humans. But servicing an electric car and a petrol car are two big differences already today. At some point, robots will learn how to change bolts and wheels. This is obvious enough, it can easily be extrapolated. Jobs for those who work with their hands will be around for a long time to come.
All this will gradually change the principles of economics, because the essence of economics is the distribution and redistribution of access to energy.
To put it very simply, when the fusion reactor starts up, there will be energy communism
Today, the difference between the poorest country and the richest is the number of kilowatts consumed per capita. Or everything can be transfered to kilowatts. If you take an American family and a Bangladeshi family, you see that the difference in energy access is several orders of magnitude different.
We take the household income divided by the cost per kilowatt, and we get the amount of energy for the average American and the average Bangladeshi. But once the conventionally infinite, free source of energy kicks in, consider that communism has arrived from this point of view. Do you need 50 kilowatts a day? - Here, take it. This is the inflection point for biblical economics (in the sweat of your face...).
The classical understanding of economics will disappear as soon as the conditionally free source of energy is available.
You don't have to earn money to keep your refrigerator running. Or heating or anything else. I highly recommend books by Vaclav Smil, very concentrated food for thought.
gg: And one last question: how can chatbots be used by media, gagadget for example?
Alexander Vityaz: There are many options: news streaming, polls, community outreach, subscriptions, invoices, ads, microlearning, radio - it's more a question of ideas than technical capabilities - for example, you can set up a GPT chatbot in a couple of hours that will let readers talk to the content of your whole site, which is a whole new experience for you and for users.
A riddle from Alexander Vityaz
The latest riddle about autonomous artificial intelligence from Alexander Vityaz: what is the last word in this scheme?