“Have you tried running that through ChatGPT or Claude?”
Ominous words increasingly heard around the office, as a growing number of tech workers feed their thoughts into chatbots and relay the output to colleagues as if it were their own.
For many, this smacks of: I don’t care enough about this to write it myself.
“I get the promise of saving time but it doesn’t half kill motivation,” a friend of mine recently told me, after receiving an AI-written instruction. “Feels incredibly inauthentic,” said another.
On dating apps, there's already a term for someone suspected of using AI to message you: a chatfisher. Reddit is full of office workers venting about their corporate equivalents.
One commenter described a colleague's email that gave the game away immediately: "This girl at my work recently sent me an email that started with the words 'Here is a polite, less angry version of your email'."
Artem Kuchukov, CEO of German robotics company Kewazo, is a founder who'll admit he fell into the practice of over-using ChatGPT to communicate with his team.
“I started personally announcing every new person who is joining our company,” he says. “When I did it for the first two times, I used ChatGPT, and it saved me a lot of time.
“After that, though, I started getting sarcastic comments from my team, such as ‘This email sounds so ChatGPT’. Basically, people saw through this very quickly and did not like it at all.
“So I had to change my approach and combine ChatGPT messages with personal touches so that it felt less robotic,” Kuchukov says.
‘The bar is getting lower’
The stories kept coming.
One salesperson at a Berlin software company described receiving a suspiciously AI-generated reply from her boss after she messaged to say she'd be out sick. “Was it really so hard to type, ‘no problem, feel better soon.’? No, this guy had to prompt ChatGPT for a response. The bar is getting lower.”
That, at least, is fairly harmless. More irksome are the cases where managers use AI to second-guess their own employees.
“I’ll show our founder something and he’ll literally say, ‘great, let me just see what Claude has to say’,” an employee at a London-based fintech told me. “I will then have to justify my work after Claude has picked holes in it.”
Claude, according to this person, is becoming a kind-of-a second boss in his workplace. A consultant few like, who springs "gotcha" questions at staff.
To be fair, not everyone is concerned about AI communication. “Honestly I'd prefer my boss not write to me at all, but if he's [going to] bug me he may as well use ChatGPT so the spelling is correct,” one worker told me.
‘You can’t automate humour’
Some bosses are in the clear. Ilan Fisher, communications lead at AI startup Wonderful, doesn’t suspect his superior of any foul AI play. “It would be hard to automate insider and niche British humour,” he says.
Jan Čurn, CEO of Prague-based web scraping startup Apify, also doesn’t use AI to write to his colleagues.
“I have the opposite problem,” he tells me. “Instead of senior staff, it's sometimes the more junior people who write meeting minutes, performance feedback, specifications or launch plans, from which you can clearly feel the AI slop.”
Some are sympathetic to the chatbot-dependent boss. The Berlin salesperson admits her boyfriend put things in perspective when she complained about her CEO. "He said, well, 'at least he's using AI. You don't want to work at a company that still hasn't discovered chatbots'."
She also confesses to deploying AI in her own client emails as a last resort. "It's a plan B when you can't find the strength to be polite to assholes."
