Analysts at Sber's Center for Macroeconomic Research have completed a large-scale study assessing the potential of implementing generative artificial intelligence in the Russian economy. The key conclusion: neural networks can add 1.2 percentage points to labor productivity growth.
This is not just an abstract figure, but the result of segmenting the labor market into four categories based on the vulnerability of professions to algorithms. The greatest effect is expected in sectors where the primary product is textual information. It is here that generative models demonstrate maximum efficiency, creating content faster and better than humans.
Who will benefit from AI implementation
First and foremost, specialists whose routine involves data processing and communications will see high returns from algorithms. These include:
- Sales managers.
- Software developers.
- Accountants and auditors.
- Marketers and PR specialists.
- Call center operators.
For example, sales managers will be able to send out commercial proposals faster, marketers will generate advertising slogans more quickly, and call center employees will get a "smart" assistant that handles standard phrases like "your call is very important to us."
Medium level of impact
An above-average contribution is expected from another category of specialists whose tasks are partially automatable but require human involvement:
- Sellers and administrators.
- Store directors.
- Engineers and cost estimators.
- Lecturers and teachers.
The situation with teachers is especially curious. Students began using neural networks significantly earlier than the education system adapted to this. As a result, learning risks turning into a dialogue between two algorithms communicating through intermediaries.
Who will remain outside the influence of algorithms
The group with low exposure to AI includes representatives of physical labor:
- Machinists, drivers, and freight forwarders.
- Fitters, auto mechanics, and welders.
- Installers and electricians.
- Loaders, warehouse workers, and packers.
- General laborers, cleaners, and waiters.
- Chefs and nurses.
A cutting-edge language model will not be able to unload a truck, weld a seam, or set up an IV drip. This work will continue to be done by humans, so manual labor productivity will remain at its current level. Doctors and couriers fell into an intermediate category with below-average impact.
Thus, the sought-after 1.2% will be provided by office workers, whose work is best suited to automation. Professions that neural networks cannot reach will continue to operate under conditions of labor shortages.
My view: Sber's assessment looks conservative but realistic. In the short term, AI will indeed intensify inequality between "digital" and "physical" professions, but this is precisely what will become a driver for revising the education system and retraining personnel. Investors should take a closer look at companies that are already implementing generative models into operational processes — they will gain a competitive advantage in the next 2-3 years.