Microsoft clarifies the scope of AI in jobs and tasks | Keryc
Microsoft published a note to clarify what their research does and doesn’t say about how artificial intelligence applies to occupations and tasks. The bottom line? The study aims to map where chatbots are useful, not who loses their job tomorrow. (microsoft.com)
What the research found
The researchers looked at how people use generative systems like Copilot and then compared those activities with the O NET database to see which occupations include those tasks. They found AI shows greater applicability in cognitive and communication tasks: writing, gathering information, teaching and advising are the activities where the tool contributes most. (microsoft.com)
Does that mean jobs will disappear? No. The team itself warns that their data do not prove an entire job will be replaced; they show potential applicability to specific tasks, not a verdict on employment. (microsoft.com)
Methodology, explained without jargon
In plain language: the authors used anonymous user conversations with Bing Copilot and labeled them by activity. Then they matched those labels with O NET to calculate an "AI applicability" score per occupation. The dataset included around 200,000 conversations and reflects real user behavior, with measurements of how well the AI performs certain activities. (microsoft.com)
That makes patterns easier to see: for example, occupations in computing, administrative support and sales show high scores because many of their tasks involve information and communication.
Key limitations you should know
No study captures the full complexity of human work. O NET lists tasks, but it does not reflect interpersonal judgment, ethics, or domain expertise that make a professional valuable. Turning an applicability score into a prediction of unemployment is a bad interpretation. (microsoft.com)
There are also biases in the data: they come from Copilot users between January and September 2024, which introduces effects of access, familiarity and purpose of use. Also, the analysis is limited to chatbots; other forms of AI are not covered here. (microsoft.com)
Now what? Practical recommendations
If you are a worker: ask yourself which parts of your day are repetitive, search-heavy, or writing-focused. That’s where AI can help you save time. Try the tool, keep notes on what improves and what risks appear (quality, bias, privacy).
If you are a manager or entrepreneur: measure tasks, not job titles. In projects that involve text, information management or basic advising, run small pilots with clear metrics (time, accuracy, satisfaction). Invest in training and in protocols for human oversight.
If you are a policymaker or in HR: the answer is not to ban or to embrace blindly. You need a mix of training, impact assessment and labor rules that recognize complementarity between AI and human work. Microsoft also suggests following broader initiatives and reports on the future of work to inform policy design. (microsoft.com)
The real news is not that AI will wipe out professions overnight, but that it is already changing concrete tasks. The important question? How do we make that change work for productivity, fairness and dignified work. Microsoft’s research provides data and clear limits; the rest depends on what we decide to do with them.
Stay up to date!
Get AI news, tool launches, and innovative products straight to your inbox. Everything clear and useful.
Microsoft clarifies the scope of AI in jobs and tasks
