Generative AI usage at the moment centers overwhelmingly on information work such as creating, processing, and communicating information, a Microsoft Research study "Working with AI: Measuring the Applicability of Generative AI to Occupations,” authored by Kiran Tomlinson, Sonia Jaffe, Will Wang, Scott Counts, and Siddharth Suri finds.
That should come as no surprise, since the whole point of generative AI is the creation of content.
Generative AI is most often invoked by workers for:
Teaching / explaining
Writing / editing
The most-frequent AI actions include:
Providing information / assistance
Explaining / teaching
Responding to inquiries
Preparing informational materials.
The study also suggests AI performs best at:
Providing information to customers/public
Editing written materials
Writing commercial/artistic content.
On the other hand, generative AI does less well for:
Physical / manual tasks.
In terms of where generative AI should affect jobs, the researchers say the highest applicability is for:
Media & Communication Workers
Sales Representatives (Services)
Information & Record Clerks
Interpreters & Translators
Writers & Authors
Customer Service Representatives
Journalists, Editors, PR Specialists, Data Scientists, Management Analysts
The lowest applicability (or job displacement or devaluation) happens in predominantly manual, physical, or direct human-care roles, such as:
Construction & extraction trades
Agricultural workers
Building cleaners
Food processing / machine operation
Many healthcare support and production roles.
The researchers examined about 200,000 anonymized U.S. conversations with Microsoft Copilot (now Microsoft Copilot) from January 2024 to September 2024 as part of the research.
No comments:
Post a Comment