“Work garbage” is flying around, and technology giants who once strongly embraced AI have begun to “press on the brakes”

IT House news on September 19, local time on the 17th, according to the Fortune magazine website, technology industry leaders who once touted AI as the key to unlocking the full potential of employees are beginning to gradually change their views.

Shopify co-founder and CEO Tobias Luedtke is one of them. Last year, he also asked employees to make the use of AI a basic requirement and to prove that existing jobs cannot be done with AI before requesting more resources.

Now, Lütke is criticizing his employeesHanding over unchecked AI emails and code without taking responsibility for rough results. “We call these ‘garbage grenades’ and people are throwing them around each other. It’s definitely a bad thing.”

Luedtke noted that it would be easy to give AI a free hand, but misusing it would only shift the problem to the people receiving and reviewing the work. For example, AI is supposed to help organize the key points of an email, rather than expand a few sentences into a long speech that wastes other people’s time. “You didn’t read it carefully yourself, but now you have to ask your colleagues to review it, and they will say, ‘This is not quite right’.You just leave the work directly to the AI ​​to do it。”

Duolingo CEO Louis von Ahn also startedRetracting previous radical stance on AI. Last year, he announced that Duolingo would shift to “AI first”, where employee evaluations would be based on AI usage, and some human contractors would be replaced by AI, adding manpower only when the team cannot automate the required work.

In an interview with “Fast Company” in May this year, he admitted that heEver been carried away by how well AI performs in writing demos, but it was only after the scale of use was truly expanded that we discovered thatAI will ultimately fall short of employee creativity. “We might need to write 1,000 different stories to help people learn a language, and at the end of the day you’ll find that 20 percent of them are complete garbage. Whenever we use AI to do a lot of work at scale, we have to be very careful not to let this garbage get in.”

Researchers call the phenomenon Lütke describes “workslop.”AI output that is exquisite on the surface and looks decent, but actually lacks practical content and needs to be reworked by others.。

BetterUp Labs and Stanford University’s Social Media Lab surveyed 962 full-time U.S. office workers this year. 52.7% of respondents said they had sent AI work rubbish to colleagues, andThe more your organization encourages the use of AI, the more common this will be. Another 38% of respondents reported receiving this type of content, spending an average of 3.4 hours per month revising it.

AI work junk is different from ordinary low-quality work done by humans, in that it looks like acceptable work on the surface.There is a lack of really useful content inside. Examples cited by interviewees includeThe format and structure are complete, but the link cannot be opened.件，以及Code that could have been very simple, but was made extremely complex by AI。

Such problems can also damage coworker relationships. Surveys show that once employees suspect they are receiving AI work spam, they view the sender as less competent and less friendly. Of those who had received such content, 36% also said they would like toAvoid working with these colleagues again。