First they were laying off, now they are hiring again. There, artificial intelligence did not work
Artificial intelligence automates fragments of tasks more often than eliminating entire roles. He is more of a tool than a full-fledged employee. This creates competence gaps and the need for people who can implement, supervise and integrate these tools with the rest of the processes.
The International Labor Organization has been emphasizing for many months that generative AI expands opportunities and helps increase productivity, rather than being a pure substitute. It automates certain activities in competitions, but not the entire competition.
At the same time The clash of corporate ambitions with economic calculations can be painful.
Check also: Small entrepreneurs against the system. “Make up or go out of business”
Artificial intelligence not as expected
This quarter's famous MIT report on “GenAI Divide” described that as much as 95 percent implementation of generative AI in enterprises there is no measurable impact on P&L, and quick returns are achieved by a small minority of projects that solve a very specific problem.
This has echoed on Wall Street and in the industry media because it runs counter to record spending on AI.
In addition, there is the cost of simple cuts. Orgvue analyzes show that for every $1. The savings on downsizing amount to an average of $1.27. indirect costs – from severance pay and unemployment insurance to lost productivity and a wave of voluntary departures. In practice many companies discover that a shortened payroll does not simplify their technology strategy, and some of those laid off will have to be recruited back anyway.
In conversations with Axios, Andrea Derler of Visier explains that AI can be a “convenient explanation” for layoffsbut not yet justified enough to replace people on a large scale. It also points to an underestimated implementation calculation, from hardware and data infrastructure, through security, to integration with processes, which should be compared with the actual possibility of automating tasks in a given role.
“Boomerangs” are back
We can see the result of this clash of expectations and reality today, when “boomerangs” are back the fastest way to fill competence gaps in AI projects. An experienced former employee understands the organization's culture, tools and data architecture, thus reducing start-up time. At the same time, the importance of hybrid roles – combining domain knowledge with the ability to work with AI models – is growing, which is also confirmed by OECD reviews. Transformation involves reorganizing work and tasks, not mass abolishing jobs.
All this doesn't mean that AI doesn't make sense. Field studies show productivity gains in well-structured applications, but they require narrow focus, meaningful integration, and investment in people. The problem is that in many companies, implementation was still preceded by a sales presentation, not a map of processes and roles. No wonder that today some management boards return to specialists they “raised” themselves, instead of patiently teaching new employees organizational nuances.
AI is still a tool with huge potential, but without patient organizational work and investment in people, it mainly turns into expensive proof of concept projects. Data about growing “boomerangs” are not so much a failure of automation as a course correction.
