- Published on
Don't credit the LLM
- Authors
- Name
- DeepHorus
A workplace trend is raising eyebrows: employees are increasingly prefacing their work with statements like "according to ChatGPT" or "I asked Claude to write this." An essay titled "Don't credit the LLM" argues this practice is both odd and harmful. The author compares the confessions to a pilot announcing their avionics software version to passengers.
The essay, which has sparked discussion among tech workers, suggests the habit may have begun as novelty or excitement. It offers four possible reasons for the behavior: giving credit where it is due, trying to impress by association, failing to review LLM output, and complying with corporate mandates. For each reason, the author responds with a mix of sympathy and skepticism.
One example highlights a worker answering a question about support requests with "according to [LLM tool], we've received 17 requests since it was released." The essay argues such a statement is unnecessary, much like a writer crediting a spellchecker for a polished manuscript. The central problem, it says, is that tools cannot be held accountable for their output.
The piece frames credit and accountability as two sides of the same coin. Since an LLM can neither take credit nor bear responsibility, crediting it dilutes the worker's own ownership of the result. This is particularly dangerous when output is only 80 percent complete and the missing 20 percent is unknown.
The author takes a firm stance: if the work is remarkable, take full credit; if it fails, take full blame. That, the essay argues, is how meaningful work has been created long before AI existed. The advice applies to both the person who outsourced thought and the one who merely feels like an imposter.
As LLMs become further embedded in daily workflows, such disclaimers may eventually disappear. But the essay's core lesson likely remains: accountability stays with humans, not tools. The debate over whether to credit LLMs is really a debate about ownership of work.
