I've published a short review of AI and work research at The Commonplace. It is an illustrative editorial pilot built around three studies. The underlying collection's publication-date window runs from January 1 through September 2, 2026, using the snapshot published on September 2.
I discuss AI-assisted task performance alongside learning, then look at coding agents and code comprehension. A hiring study examines AI-led interviews in a setting where people retain hiring authority. The discussion keeps the studies' settings and comparisons separate.
For 2027, I propose questions about whether learning lasts after assistance ends and whether results carry across workplaces. I also want to know how changes in tasks affect pay and careers, including what happens to workers who have to move into different jobs. I chose these questions to follow up on the examples in the pilot.
The review also explains some limits of the collection behind it. Automated topic tags can help locate papers, but they cannot establish what researchers have or haven't studied. A collection date records when a paper entered the system and can differ from its publication date. This is a selected reading, not a systematic review of the year's literature.
I'm publishing the written review without the annual evidence map or bulk data downloads.