An algorithm took the commission, and an AI course gave the work back: the story of Ira and Max

Positions that call for AI skills pay on average 56% more than comparable roles without them. That is the conclusion PwC analysts reached in the Global AI Jobs Barometer 2025, after analysing almost a billion job adverts. The same report found that the number of vacancies requiring such skills rose by 7.5%, while job adverts overall fell by 11.3%. This is a global average, and in Ukraine the figures will be more modest. The direction is clear all the same: the market pays for people who know how to work with the algorithm rather than compete against it.

Ira, a translator from Kharkiv, was caught off guard by these statistics. She had started 2023 in Dnipro with a steady stream of clients, and by the end of the year agencies were sending her “drafts” that needed only a light polish and paying her proofreading rates. There was nowhere to work at home: she shared a single laptop with her partner, and the internet disappeared whenever the power did.

Max, an administrator at a small office, was thinking along different lines. He entered records into spreadsheets and replied to the same messages over and over. He did it well, which was exactly why he knew such tasks would be the first to be automated. There was another complication. His documents still didn’t match the name he goes by, so every interview began with a pause he had learned to sit through. He was looking for remote work where people judge results rather than an application form, but a portfolio had to be built from something.

Both were brought to the AI course by an advert from ALLIANCE.GLOBAL. The organisation works with the LGBTQ+ community, helping with shelter, humanitarian aid, and psychological and medical consultations. The course ran online, in Ukrainian, with trainers and a group of people who didn’t need it explained why an application with the “wrong” name never got any further. Ira was given a laptop by a community centre, and mobile internet to cover the power cuts was arranged separately.

For the final practical task they were paired up. Max looked for ways to automate replies to clients, while Ira corrected the places where the model was confidently wrong. There were plenty: invented terms, wrong grammatical genders, and dry officialese where natural, living language was needed. That was when she changed what she offered to clients. Instead of “translation”, Ira began proposing localisation with an AI draft and a human proofread, backed by a glossary of terms and fact-checking. She won back two former clients and found another on her own.

Max put together a portfolio of three pieces: client reply templates, automated schedule planning, and sorting enquiries by type. His first conversations with employers went through a test task rather than an application form. He now works remotely as an assistant operations manager at an online shop.

There is no magic in any of this. Ira still checks every paragraph, and Max keeps having to show that a template is no substitute for attentiveness. The course didn’t promise them a job, but it did give them a tool, a laptop and a group in which they could develop.