A new study shows that it’s leading to an overwhelming amount of low-quality output labeled “workslop,” which is creating more work for people on the receiving end and costing businesses money. Researchers from the Harvard Business Review, in collaboration with the Stanford Social Media Lab, coined the term “workslop” to describe the emerging problem.
HBR’s survey of 1,150 full-time employees across industries in the US revealed the pervasiveness of workslop:
- The study said it disproportionately affects the tech sector. But 40% of respondents overall say they have received workslop in the past month.
- Those who report having had workslop poured onto their desk said an average of 15.4% of the work they receive fits the slop description.
- While workslop is mostly a peer-to-peer problem (40%), it comes to managers from direct reports 18% of the time.
The emotional and social toll: When asked how it feels to receive workslop, 53% of employees reported being annoyed, 38% feeling confused, and 22% feeling offended. Along with having to scour the slop for mistakes and missing information, recipients feel the stress of having to discuss the issue diplomatically with peers and bosses.
The bottom line: The study estimates that for a company of 10,000 employees, workslop results in $9 million worth of lost productivity per year. The HBR study echoes one from the MIT Media Lab in July that found that despite between $30 billion and $40 billion being invested in gen-AI tools, 95% of organizations are getting zero return from the technology
— * * *—
Konsulentfirmaene har doblet og tredoblet sine overskudd, blant annet gjennom å utrede hvordan andre firmaer kan nyttiggjøre seg AI. Samtidig er jobben de utfører en av de som lettest kan erstattes med AI: Sluke store datamengder og destillere det til en glanset Powerpoint-presentasjon.
Harvard Business Review «AI is changing the structure of consulting firms»: Færre juniorer på gulvet, det samme antall seniorer på toppen.
Fortune: Deloitte caught using AI in $290 000 report to help the Australian government crack down on overly generous welfare, after a researcher flagget hallucinations. AI output had not been fact checked. At the same time Deloitte proudly announced the were investing $3 billion into AI.
Fortune: MIT reports 95% of generative AI pilots at companies are failing.
University of Cambridge: How will AI change the way we conduct scientific research?
Harvard Business Review: AC Can (Mostly) Outperform Human CEOs: When researchers at the University of Cambridge pittet 344 human competitors against a leading off-the-shelf LLMs, the chatbot beat the top participants on almost every metric. It also was fired more quickly. The simulated an automative company where the participants had to take key business decisions. Humans were also making hallucinations. Additionally, the LLMs hade a far greater availablility than human executives.
The main challenge is that you can’t send AI to prison, if the company breaks the law.
The research only applied to tech firms, this may be used as an excuse to combine multiple roles in one person and the AI was bad at handling black swans. Business leaders also consider their own jobs as far more diificult to replace.
Fordelen med å erstatte folk med AI og roboter fra toppen og ned: Disse folkene er de dyreste. En amerikansk CEO er verd 250 arbeidere. Det er dyrt å investere i utvikling og drift av en skreddersydd AI-modell, men du kan bruke 20 datakonsulenter på å drifte en CEO og fortsatt finne høy lønnsomhet.