1 hour ago
1
July 31, 2026 — 5:01am
I used to have a strong professional relationship with my boss and I respected them. But they have changed recently. Once they were considerate and supportive – some colleagues even thought they were a bit too gentle. Now they are neither of those things. They are definitely not gentle. They have started talking about instituting a new regime based on the advice of a friend and business mentor. All the new policies and ideas seem severe.
A small group of people who report to this boss (including me) have talked in private about who this “mentor” is and why they are telling our manager to be so harsh and punishing. Our team is performing well, and has for years. There is now a strong rumour that the friend is actually an AI chatbot. Surely, that can’t be true, can it?
It certainly seems extraordinary that a professional could become so attached to a chatbot that they would change their entire work style and personality based on its “opinion”. Extraordinary but not inconceivable.
About three years ago, a reader wrote to me about a work problem that they had reported to their HR team. They had been rudely brushed off. When they experimentally asked a chatbot about the same problem, the AI response was infinitely more empathetic and sensitive.
Although we can put this down to the tendency of modern large language model (LLM) chatbots to validate and endorse the concerns or ideas of its user, partly as a way of keeping the conversation going, an unflattering term like “AI sycophancy” doesn’t change the experience of that reader. Or anyone who speaks with AI and finds “someone” who appears to genuinely listen.
I’m not going to say I know for sure what is going on with your boss. They may not be consulting a chatbot at all (although I find the evidence proposed in your longer email, via your worried colleagues, compelling). But to answer your direct question: yes, it can absolutely be true.
People today are anthropomorphising AI in all sorts of different ways – this is no longer a fringe practice.
Some people develop what are known as parasocial relationships with this new generation of ultra-coherent AI models. “Parasocial interaction” is a term coined in the 1950s to describe, for example, the way a TV viewer might come to think of a media personality as their friend, despite the relationship being entirely one-way.
Now, some experts in the field believe “parasocial” is not quite the right term to apply to profound relationships between humans and AI. Whereas in the ’50s a TV presenter might inadvertently (or at least perfunctorily) encourage parasocial intimacy by looking down the camera and talking directly to the audience, in the 2020s a chatbot mimics interest, understanding and even love in a highly sophisticated, often utterly earnest, way.
Let’s stick with the term for now, at any rate.
We know that people today are anthropomorphising AI in all sorts of different ways – this is no longer a fringe practice. They converse with their favourite chatbot as they might a human therapist, a companion, a “troubleshooter” and even a “work buddy”.
If your boss is doing that, but has entered into a pathological version of parasocial engagement, it’s quite possible that they’re taking advice from this digital confidant in good faith, genuinely understanding it to be their intimate. And perhaps their teller of cold, hard truths.
There could be any number of reasons for how this has come about. It might have been a conversational chain reaction, beginning with a benign question but branching, exchange by exchange, into something much more radical.
Or it may have been less accidental. Perhaps your boss, worried that they were seen as a soft touch, sought advice on changing this perception, and were given a menu of aggressive and ruthless management tactics.
The change won’t have come out of nowhere. And while I think it’s sensible to be curious about, and even sympathetic to, what catalysed it, you don’t need to tolerate or excuse the cold-hearted turn.
Your manager, of course, has the prerogative to make changes, and is under no obligation to be your friend, but if they’re taking dogmatic advice from a chatbot, I think they’re engaging in an ill-advised, even dangerous practice – and possibly stepping onto legally shaky ground.
One day in the future consulting with robots as if they’re trustworthy humans might be entirely normal, and possibly beneficial. That day has not arrived.
Send your questions through to Work Therapy by emailing [email protected]
Get workplace news, advice and perspectives to help make your job work for you. Sign up for our weekly Thank God it’s Monday newsletter.
