Global studies released on the same day are throwing cold water on one of the most popular narratives about artificial intelligence: the idea that it will wipe out jobs.
The reality the data shows is quite different.
Instead of replacing workers, AI is behaving more like a collaboration tool — one that helps you think, organize, and find information, but still relies heavily on humans to actually work. 🤝
Two independent studies, both published on Thursday (July 23), reached similar conclusions and reinforce this point. One comes from Google, the other was commissioned by the Adecco Group, a Swiss company specializing in talent and technology. Together, they cover everything from how people use AI on a daily basis to the real impact of the technology on the job market across dozens of countries.
What they found might surprise anyone who still believes automation is going to sweep entire professions off the map.
What the data actually says about AI and jobs
Google unveiled its first study called the AI & Economy ATLAS, which stands for Activity, Task, Landscape and Adoption Study. The main takeaway sets the tone right away: artificial intelligence shows up in 68% of occupations, but is used in only about 21% of the tasks within a typical job. In other words, the technology is spread across nearly every sector, but it is far from dominating everything a person does during the workday.
To arrive at that number, Google analyzed no fewer than 15 million interactions — aggregated and anonymized — between humans and AI, collected across its products and tools. That massive volume of data reveals real usage patterns, not just assumptions about how people might be using the technology.
And the pattern that emerges is clear: the tasks where AI gets called upon tend to be part of a collaboration, not full-blown automation. We are talking about things like brainstorming ideas, defining strategy, searching for information, and learning. Less than 10% of work-related AI interactions fully automate a task. That changes the conversation quite a bit, right? The machine helps, suggests, and speeds things up — but the person making the final call is still a human. 🧠
The Google report also revealed a data point worth paying attention to: only 3% of occupations showed AI usage across more than 75% of their tasks. That small group includes professions like software quality analysts and testers, human resources specialists, and document management specialists. These are exceptions, not the rule. For the overwhelming majority of roles, AI remains a targeted assist within a routine that still depends heavily on human judgment.
The global job market picture
The research commissioned by the Adecco Group reinforces this reading from a different and complementary angle. According to the company, the central finding is that AI is changing tasks faster than it is eliminating jobs. And the numbers backing up that claim are pretty compelling.
After three and a half years of ChatGPT being available to the public, employment rates remain at record levels across 38 member countries of the Organisation for Economic Co-operation and Development, commonly known as the OECD. On top of that, employment in roles with direct exposure to artificial intelligence has held steady. No collapse, no mass layoffs driven by the technology.
And here is a detail that runs completely counter to the end-of-jobs narrative: 1.9 million new AI-related jobs have been created. That means the technology not only failed to destroy jobs on a large scale but also generated unprecedented demand for professionals who know how to work alongside it.
Another interesting finding from the study is that fewer than 10% of companies in the United States have integrated AI into their core workflows at scale. That means even in places where the technology is most advanced, adoption is still relatively modest. And where it has already been deployed, the primary effect is reshaping tasks, workflows, and skill requirements — much faster than it is eliminating entire occupations.
Denis Machuel, CEO of the Adecco Group, summed up the current moment well. According to him, the next phase of AI will be won by those who know how to redesign work — understanding what people should do, what AI agents can do, how skills are evolving, what competencies will be needed in the future, and how value is created and measured. That is a far more strategic vision than simply swapping humans for machines.
Collaboration as the model that is actually working
One of the most interesting findings from the studies is that the collaboration model between humans and AI is proving far more effective than the total replacement model. When workers have the autonomy to decide how and when to use artificial intelligence tools, the results are consistently better than when the technology is imposed from the top down, without adapting to the context of each role.
This makes complete sense when you think about what AI actually does well: it excels at processing large volumes of information quickly, identifying patterns, and generating initial drafts of content or analysis. But it frequently stumbles in contexts that require cultural sensitivity, situational ethics, or simply the common sense accumulated through years of hands-on experience.
This point is especially relevant for knowledge workers — analysts, writers, developers, healthcare professionals, and educators. Across all of these fields, the data shows that AI works best as a thinking partner rather than a replacement. A doctor who uses AI to cross-reference symptoms against medical databases still needs to look at the patient, listen to their story, and make a decision that factors in things no algorithm can fully capture. A writer who uses AI to structure ideas still needs their own voice, cultural repertoire, and the ability to emotionally connect with the reader. The technology supports, but the final decision — and the responsibility for it — remains human.
This pattern repeats itself across sectors ranging from retail to engineering, and that is where a significant mindset shift is emerging from the data. The organizations reaping the most benefits from artificial intelligence are not necessarily the ones that automated the most processes, but rather the ones that built a culture of genuine collaboration between teams and technology. They invest in training, encourage experimentation, listen to workers about what works and what does not, and treat AI as a resource that needs continuous calibration — not as a magic solution that fixes everything on its own. 🚀
Why this debate matters right now
We are living in a moment when decisions about how to adopt artificial intelligence are being made in real time — by companies, governments, and individuals alike. And the narrative dominating this debate has a direct impact on those choices. If the prevailing idea is that AI will inevitably eliminate jobs, it can lead to both exaggerated resistance and hasty corporate decisions to lay off teams before understanding the real impact of the technology.
The global studies published this week are a necessary counterpoint to that cycle, because they put concrete data where speculation used to be. This is not guesswork: it is millions of real interactions analyzed by Google and data from dozens of countries compiled by the Adecco Group.
Understanding that automation has clear limits — especially when it comes to tasks involving complex human judgment — is not naive optimism. It is an honest reading of what current AI models can and still cannot do. Large language models, as impressive as they are, do not have consciousness, do not understand emotional context the way a human does, and make mistakes that any experienced professional would spot immediately. That does not diminish the value of the technology — on the contrary, it helps position it more intelligently within work processes.
It is also worth remembering that this transformation is not exclusive to higher-paying roles. Recent analyses already indicate that AI and automation are advancing beyond high-salary positions, reaching blue-collar and operational occupations as well. This reinforces the importance of thinking about professional reskilling broadly, reaching workers at every level — not just technology specialists.
The takeaway from these two studies is straightforward: artificial intelligence is one of the most powerful tools the job market has ever seen, but it needs humans to make sense. Collaboration is not a Plan B while the technology catches up — it is the model that is working right now, in real data, at real companies, with real workers. And those who understand this sooner will come out ahead — not because they will use more AI, but because they will know how to use it better. 💡
