Evgeny Burnaev also said that for now the systems

[tanjimajuha20]

Worlds?"

"We have a set of technologies that generate models for data scientists quite well - 100-150 times faster than a person does. However, in reality, projects are accelerated three times, because all the options that it has made have to be checked. The more work we give to not very qualified artificial intelligence, the more effort we spend on checking it," said Alexander Bukhanovsky.

Head of the Teachable luxembourg whatsapp resource Intelligence research group, AIRI professor, and director of the Skoltech AI Center Evgeny Burnaev noted that automation allows for more efficient work and more projects to be carried out simultaneously. However, he agreed that AI errors are one of the challenges in implementing the technology: "Assistant systems like Copilot are emerging that really speed up the work of employees, but this is a double-edged sword. I have heard colleagues say that with the help of Copilot you can do something that you then have to hire additional people to fix it. Therefore, the effect is still being assessed."
require the involvement of professionals so that they can evaluate the results of the AI’s work.

The moderator of the discussion, Andrey Kuznetsov, director of ML at Odnoklassniki and VK, recalled cases where organizations conduct code reviews, and a large language model tells them what needs to be corrected. In this regard, the question arose: won't it turn out that people will trust AI decisions too much and approve them without checking, which will lead to errors?

"The problem of a person trusting AI and being lazy is no different from a person simply being lazy and not checking. If this happens, we will have to create a control system - another language model that will tell the specialist to check for AI," said Evgeny Burnaev.

Experts have come to the conclusion that AI does not have reflection, intuition, or insight - so far only humans have this capability, so the technology will not be able to replace a specialist in the near future.

"Artificial intelligence helps a person analyze an order of magnitude larger volume of data, which a person was not capable of doing before, and find something in it that will lead to insight. But if in the future AI gains the ability to reflect, it will be very interesting," says Evgeny Burnaev.

Read also

Smart bot reduces OTP Bank customer service time by a third
OTP Bank automated 54% of chats with clients. Due to this, service time was reduced by 30%. Also, the implementation of the chatbot allowed maintaining the level of service during peak periods.

Nevertheless, AI will develop rapidly and will be actively used by people in their work. The head of the Gazpromneft-CR partnership development program, Alexey Tisenkov, predicted the growth of AI-based assistants. He also believes that they will soon replace entry-level specialists. He spoke about this at the session "Assistant Professions Based on Artificial Intelligence".

Alexey Tisenkov recalled a study by Bain & Company, according to which in the future there will be significantly more assistants who will replace standard personnel: "Most likely, these will be "baby" programs that will replace juniors. After graduation, current students will compete with assistants, of which there will be many. In this case, it would be good to quickly become a team leader or a senior-level employee."

He also believes that in the future, anyone will be able to choose an assistant from a store who will specialize in the required area - it will look something like an exchange for finding orders.

"Various organizations have begun to build similar systems; last week we talked with EPAM, they also began to build an open-source solution that allows connecting various LLM models to solve problems that they have in production," said Alexey Tisenkov.