'Workslop' is the latest anti-AI term and leading AI academics warn it could be more widespread than initially thought
AI “workslop” is clogging up businesses as “low-effort, passable-looking work” is handed around for others to fix up, correct and re-interpret, potentially costing businesses millions every year, according to a new report.
“Workslop” is a newly coined term for AI-generated work that “creates the illusion of progress – slick slides, lengthy reports, overly tightened summaries, or code without context,” according to the new report from Stanford Social Media Lab and BetterUp Labs published last week.
“Rather than saving time, it leaves colleagues to do the real thinking and clean-up,” the report said.
Workslop is having an “insidious effect” throughout businesses - shifting work up or downstream, with workslopper’s colleagues left to correct mistakes, add context, interpret confusing content or simply re-do work.
Workslop floods businesses
The report found workslop is flowing freely around organisations, with most passed between colleagues (40%); direct reports ‘workslopping’ up to their managers (18%); and managers slopping down the chain (16%).
Workslop is also time consuming and costly, with the survey of 1,150 US desk workers finding some 460 received workslop in the last month.
Each incident is taking up to two hours for each worker to resolve - at an average cost of US$186, which could cost a 10,000-person company US$9 million annually - or AU$13.6 million annually.
Worksloppers seen as less intelligent, creative capable
Receiving low-energy workslop is also proving awkward for colleagues to negotiate, with 53% saying it made them annoyed, 38% saying it triggered confusion, and 22% saying it offended them.
Furthermore, around half of respondents saw their workslopping colleagues as less creative, capable, and reliable, 42% viewed them as less trustworthy, and 37% saw them as less intelligent.
Many people have a growing sense that AI is “happening to us” without workers being consulted, Lauren Perry, Responsible Technology Policy Specialist at UTS’s Human Technology Institute told HRD.
The rise of derogatory terms like “clanker” and workslop may be part of individuals recoiling from the rapid adoption of AI.
“The rise of this terminology is a way of people being able to push back and kind of reject this feeling like it's happening to us, and we don't get a say,” Perry said.
“It's almost like it's happening around us, and we're all being taken on this ride.”
Workslop a ‘more pervasive' issue for businesses
“This isn't a unique problem. There's always been sloppy work, or people just not putting in the hard work,” Perry said.
However, workers who previously fell behind due to being untrained, or just a bit junior, could easily be coached as they matured in a workplace.
“This problem with workslop is much more pervasive,” Perry said.
“It's the sheer volume of it which means that actually it's quite hard to deal with.
“Actually having workplace training and bringing people along on an AI journey is really important.”
Extent of workslop potentially much higher
A lot of workplace AI is likely slipping through undetected, and the true extent of workslop could be higher than the Stanford report suggests, Professor Mary-Anne Williams, Founder and Lead for the UNSW Business AI Lab and Deputy Director of the UNSW Institute for the UNSW Business School, told HRD.
“If you've detected some workslop, then you're probably detecting less than 50%,” Williams said, “because we just let things go through if it looks reasonable, especially if we’re time poor.
“We are all fooled by the reasonableness of generative AI.
“It sounds reasonable if you don't have time to actually think about the detail, or have time to check on the numbers.”
A ‘symtom' of AI's deeper issues
Workslop is the latest manifestation of a larger issue created by AI - the technology’s potential to stir up thousands of non-urgent tasks for businesses, directing resources away from high-impact work, Williams said.
“Workslop is a symptom of a much deeper problem,” Williams said.
“People have been questioning the cost of employees using very powerful tools like generative AI because they could be creating work that doesn't need to be done,” Williams said.
“It's not just how to use the technology, it's what to use it for.
“That is actually the deeper question, and that requires training in innovation - not necessarily (in) the technology.”
Australian businesses are encountering a massive gap in training employees in how to use AI to find and prioritise problems within their business that are actually worth solving.
“That's the biggest predictor of success,” Williams said.
“Because if you find a problem that isn't worth solving, and direct resources to solve a problem that's not worth solving, then that is just pure waste.”
