A personal essay titled “I’m tired of talking to AI” described a breakdown in communication when people forward machine-generated answers without checking whether they address the question. The writer’s objection was not simply to chatbot use. It was to interactions that appeared human but transferred judgment and attention to an automated system.
One incident began after the author found GitHub repositories that appeared to distribute malware. An AI service did not provide useful guidance, so the writer opened a discussion on GitHub seeking help from other people. A respondent posted the same wording the chatbot had already produced. After the writer identified the duplication, that comment disappeared. A second response then repeated the AI output again.
The essay presented this sequence as a failure of a community support channel. The author had moved from software to a public discussion because the automated answer was inadequate, only to encounter people routing the request back through the same type of system. The evidence contains the writer’s account but does not identify the repositories, respondents or malware, so those details cannot be independently assessed from the packet.
A workplace exchange produced a similar complaint. While employed as a developer, the author asked a business owner about a business task. The reply was a screenshot of a ChatGPT answer. The writer said it did not relate to the original question and contained incorrect information. Within a minute, another screenshot arrived, which the author interpreted as evidence that the owner had forwarded output without reading it carefully.
The third example involved Reddit. Someone contacted the writer about a post and continued exchanging messages. Only after several replies did the author conclude that the other side was an AI agent. The source does not explain how that determination was made, so it remains the writer’s perception of the conversation.
Across the three episodes, a consistent concern emerged: apparent responsiveness can conceal the absence of engagement. A pasted answer may arrive quickly, yet fail to incorporate the questioner’s context, evaluate accuracy or acknowledge that the first attempt was unhelpful.
The commentary offered no survey data and did not claim that every AI-assisted response behaves this way. It documented one person’s experiences across an open-source platform, a job and a social network. Its request was straightforward but increasingly difficult to guarantee online: when seeking help from another person, the author wanted that person to read, understand and answer rather than silently substitute a chatbot. The piece therefore framed disclosure and attention as central issues in AI-mediated communication.
