Time travellers are using LinkedIn to teach us about artificial intelligence

{"text":[[{"start":0.5,"text":"NBER:"}],[{"start":null,"text":"

"}],[{"start":1,"text":"FTAV:"}],[{"start":null,"text":""}],[{"start":9.64,"text":"Disappointingly, the recently published NBER paper “Time Travel on Professional Profiles” is not as sci-fi-y as its name might suggest."}],[{"start":18.42,"text":"Authors Nicholas Bloom, Gideon Moore, Lisa K. Simon and Caelan Wilkie-Rogers have not been hunting down self-proclaimed time travellers. Instead, they’ve used cached data to monitor how LinkedIn users “retroactively edit the title or description of a job they have already left”."}],[{"start":34.52,"text":"The thesis of the paper hinges on the idea that while a static assessment of mutable profiles is inherently weak for understanding past trends, one that follows the mutations reveals more."}],[{"start":35.02,"text":"As they write:"}],[{"start":45.24,"text":"These “time-travel” edits are closely tied to labor market transitions: around such edits, workers are much more likely to change employers as compared to later-editing users. This mutability can bias historical measures of skills, but it also reveals workers’ beliefs about which skills are in demand."}],[{"start":45.74,"text":"[…]"}],[{"start":63.08,"text":"Our method requires only that workers believe the skills on their profile influence their hiring probability; we have nothing to say about the efficacy of these revisions in job search."}],[{"start":72.64,"text":"In a sense, this provides a “supply side” counterpart to the “demand side” measurements of changing language in job descriptions. Paper:"}],[{"start":78.92,"text":"Time travel holds the underlying job fixed, making changes in wording likely to reflect changing perceived demand rather than new experiences. The same worker is describing the same job they held previously, where they used the same skills both before and after changing the description. The only thing that can change, then, is the way the worker chooses to describe the job: a reflection of changing demand conditions…"}],[{"start":102.84,"text":"Because time-travel is generally resume polishing, words are almost always net added."}],[{"start":108.18,"text":"To the findings. Readers who’ve witnessed the labour market themes of recent years won’t need much hand-holding on the key changes: robots up, woke down. “Established” LinkedIn users (100+ connections) who made retroactive edits to the description of jobs they’d already left between 2020 and 2026 tended to play up how much they used artificial intelligence tools in previous roles, while de-emphasising DEI:"}],[{"start":null,"text":""}],[{"start":108.68,"text":"(Zoomable)"}],[{"start":132.4,"text":"How widespread are such edits, and how far back do they go? The researchers reckon at least one in five LinkedIn users has “time-travelled” in this way and find that most edits are to jobs left years before:"}],[{"start":null,"text":""}],[{"start":145.68,"text":"The broad strokes of which types of workers are doing this are unsurprising. The most represented group is AI product managers, the least is dentists."}],[{"start":155.84,"text":"Modelling mentions of AI in LinkedIn users’ old job descriptions “as it was” (green line below) versus “as it is” indicates that users are emphasising usage more now than they did at the time:"}],[{"start":null,"text":""}],[{"start":167.56,"text":"These transitions, as you might intuitively expect, tend to coincide with users changing companies: after all, if you’re moving jobs within a company they are probably going to care less about how you describe your current job."}],[{"start":180.24,"text":"To an extent, this speaks to LinkedIn’s inherent weirdness: people are encouraged to present themselves like an item in a shop window even if they don’t wish to be bought."}],[{"start":189.76,"text":"Somewhat predictably, the paper also finds a rise in users deploying AI to edit their old job descriptions — with usage rising inversely to education levels. Also somewhat predictable is who is bucking that trend:"}],[{"start":203.28,"text":"Generally, using AI tools to edit your resume declines with education; however, MBAs exhibit significantly higher rates of AI use relative to other master’s degree holders. This association with MBAs is driven largely by relatively low-ranked programs."}],[{"start":224.3,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1786135075_1153.mp3"}