For the last 12 months, the world's media has been full of reports about Chat GPT capabilities. A neural network successfully passes tests, exams and technical interviews for a number of professions, copes with typical tasks for SEO-copywriters, technical writers, and analysts. In particular, the neural network recently successfully passed a technical interview for Google job seekers, as well as did a doctor's exam. Representatives of many professions, including IT, had bad thoughts about the need to change careers in the future and that Open AI did what the young lawyer from the bayan anecdote did. I'll try to put my thoughts on this as someone who often uses Chat GPT.
If you look at the situation soberly
For now, at least in the next 7 -10 years Chat GPT can not fully replace programmers. Yes, a neuronet can write code. Certain tasks can be assigned to it and it forms the solution in the form of a code. Moreover, it can do it in Python, Java, Javascript, assembler and even in various exotic languages like Brainfuck. And now a neural network working in test mode can be instructed to write basic things.
For example, you can delegate small games to the bot, if you properly set the task, i.e. explain what should be on the input and what should happen. To delegate some small logic. And the system will write this code, it will even work somehow. Perhaps some solutions will be quite difficult for a human. But it's important to understand that at the level of solving a business problem, it's just a piece of code that can help the programmer or business analyst speed up the task, but not do everything for them.
Often the system does it differently and not always fully, but the correctness of the responses is a matter of time. For now ChatGPT can be called more like an assistant, which speeds up the work. It can help you find bugs in the code which was "fed" to it, it can suggest a non-trivial solution, it can tell you how to write something. It is not necessarily that the hint will be correct and every single bug will be found. Also the system can create detailed documentation for the code with a description in human language about which part is used for what.
So ChatGPT can now replace some simple autotests and speed up developers' work by suggesting solutions which would take days or even weeks of work to find without it. To put it in an exaggerated way, ChatGPT is already able to take a job away from a coder-junior, but the system is still far from a middling or an experienced architect.
The problem of knowing the platforms
When creating a commercial product, it is important not only to know the language but also the platform. ChatGPT knows far not all the platforms, for example, if you take the nocode and locode platform elma 365, then asking the neural network to write an autotesting system for such a platform will not work. Because the system does not know how to write code for that platform. It takes a long time to learn, and since this platform is hardly a priority for Open AI, it is unlikely that the neural network will master such a capability any time soon.
In commercial solutions and systems integration
When we talk about large enterprises, applications are usually solved within some platform, such as SAP or 1. In order to write code for them, you need to know how they work. In these cases, ChatGPT can not in principle replace the programmer. Rather, it can write basic code, but not make it a working application. So, again, we are saying that a neural net can ease the work of a programmer, but not replace him/her.
Meanwhile, there are specialists in our sphere, which can be already called representatives of professions which are gradually dying out. For example, technical writers. So, ChatGPT is already able to cope with writing TOR according to state standards, and in general is able to generate decent enough project documentation. In addition, a neural network will take away the work of a significant number of business and systems analysts, in particular, the specialists who are primarily engaged in formalizing the requirements for ready-made templates of artifacts analysis. And those analysts who remain in the profession, i.e. capable of identifying and generating requirements, on the contrary, can significantly increase productivity by delegating routine tasks of formalizing ChatGPT.
Obviously, call center and customer support specialists working on template scripts can gradually find new jobs. My observations when working with ChatGPT show that the possibility of targeted communication on the typical customer issues, the neural network is solved more effectively than the low-skilled girls in the outsourcing call center.
In the distant future, 10 years or more, ChatGPT or similar systems are likely to become a kind of neural network orchestrator, capable of combining the automated development functions of at least simple products into a single process. It's only after that point that IT professionals in most specialties should worry about changing jobs. For example, if a website is needed, Mid Journey, will take over the visual part of the product, some other AI will create a front end based on it (so far the Open AI product is not doing too well), and ChatGPT will write the basic code, texts and generally generate data to manage the process.
Futuristic predictions
If you look a little further, in the perspective of 10 - 20 years, you can see how rapidly ChatGPT evolves. It will be constantly being "fed" more and more new platforms, examples of good solutions and well-written code, examples for specific business problems. Eventually, it will bring the system to a state where it can handle more and more non-trivial tasks. At some point in time, the neural network will be able to generate basic code directly from the terms of reference, without a lot of other additional conditions that are needed now. I assume that such a specification will have to contain inputs, outputs, expected results and data types.
Obviously, in the future with such a ToR you can get ready pieces of code, complete modules which can be combined into a complete commercial product. But the process of such integration will still require the programmer's participation. I believe that this is quite soon future, because purely technical tasks, where the developer is not required to empathize, take into account ethical criteria, will be perfectly solved by AI in 2 - 3 years. But, of course, a neural network capable of it will not be able to replace human intelligence in product development.
In the near future, neither ChatGPT nor any other neural network will be able to fully understand the task that needs to be solved, and therefore create code based on even detailed abstract tasks. In other words, the product creation will be at least tied to the work of architects, programmers and analysts. At the same time in a more distant future, let us say in the next 5 to 7 years, depending on the intensity and quality of training, a neural network may be able, with the right request, to create simple turnkey commercial products by itself on the template user requests. Perhaps its ability to generate such products will still be more modest than that of live developers.
New professions
ChatGPT will not only take the work away from people, but will create a new market of professionals. Once lowcode and nocode appeared and today they have already started to form whole teams, which work with platforms, mastering them much faster than classical programming languages. The new formation of IT professions in nosode will be ChatGPT query specialists and neural network trainers. Obviously, many of the really valuable members of the professions that will "kill" neural networks will become their trainers. They will only need to learn transfer methodologies.
Doctors have nothing to worry about
The notorious passing of the medical exam by a neural network hardly threatens the profession in any foreseeable future. As in the case with programmers, a neural network, on the contrary, for the time being only aspires to the role of an assistant, helping the practicing doctor to make the right decision and preventing possible medical errors.
The degree of risk, legal restrictions, and issues of personal responsibility will keep medicine conservative enough for a long time to prevent ChatGPT from making independent decisions about the health of living people.
This is precisely the case when the risks associated with the Lem barrier become incommensurate with the possible benefits. Meanwhile, it is already clear that in the hands of a good doctor, ChatGPT will be a tool for minimizing the human factor in diagnosis, choosing treatment tactics and assessing changes in conditions. It is important, however, that those who believe in the power of AI not become self-medicators and ignore traditional doctors, so the medical background of AI requires ethical regulation.
The cost of humanization
When it comes to such bold assumptions about the future of AI, questions always arise about the cost of modeling a humanoid approach in training a neural network. Recreating such products of the human psyche as empathy, some semblance of imagination and humor, as well as everything that we refer to as soft-skills, requires a truly cosmic amount of data. For individual companies, even large ones, the cost of training may be incomparable to the profits of replacing human functions. Only developed countries, transnational corporations, and large companies with enormous amounts of data have the resources available. It is unlikely that such an expensive solution can exist only as a product for internal use. Such a development will be profitable only if it is offered on the global market.
Open AI is counting on large-scale development. Now they rake in data from all available sources and feed it to ChatGPT, while hundreds, if not thousands of employees with sad faces from a poor African country sit and turn the chat-bot into a "neural network with a human face", preventing the appearance of what humanity fears since the appearance of the Terminator movie franchise. And the commercial solution, where these solutions are used, will apparently be offered by Microsoft. If we talk about Russia, Sber may go for such things, for example, because they have an incredible amount of data, Yandex, a large federal telecommunications operator and a large ISP. Other companies obviously won't be able to afford emulation of human-like features and soft-skills for any foreseeable future.
A little bit about Copyrite
ChatGPT, a model trained to predict the next word, used many terabytes of code from Reddit, GitHub, and Quora in training. The question arises - can the AI generated code be considered copyright free? After all, to a greater or lesser extent, it will repeat someone else's code. The corporation that has the rights to this code, within the laws of its jurisdiction can start protecting its right to this code. Demand changes, compensation for lost profits, or penalties in the form of fines.
The question arises: how can the situation be solved in the legal field? After all, the programmer who will use parts of the code generated by ChatGPT, could thus expose his company to the sword of Themis for copyright infringement. It would be hard for a company in this situation to prove that the code was generated by ChatGPT, and that the copyright was not knowingly infringed. To many, this scenario may seem absurd, but in the reality of corporate disputes copyright claims on the code are not uncommon, just remember the litigation Niantic and Wargaming conflict with Ex-development team, Press Fire Games studio (better known as BlitzTeam).
As a conclusion
Summarizing the written above, we can conclude that the programmers have nothing to fear, so all the long-term forecasts that involve the replacement of specialists with a neural network from the middle level do not have a clear time horizon and are expected no earlier than 10 years. So far, the risk zone includes technical writers, low-skilled juniors, novice analysts, and first-line customer support fighters. It is still too early to worry about professions that require systemic thinking, empathy, imagination, and creativity, i.e., creativity and characteristics that are characteristic of a living person.
