台媒台军自导自演大陆攻台 自嗨火力全面压制解放军

Question

For some reason, a culture has developed at my company where people are almost exclusively sending their requests, emails, instant messages and other communication using LLM generated text.

We are a technical team and we get these messages from the business / non tech side.

For example, I'll get many messages a day that say something to the effect of "Hey x, when you have a minute, could you review the below and answer any questions it contains?".

Note that they say when you have a "minute". But they then send a Chat GPT wall of text that's hundreds of words long that would take more like an hour or two to dissect and ascertain which parts are correct and which parts are not and then answer or clarify any questions it contains.

It's driving me nuts to the point of resignation. The sensible answer is to communicate this to them, I have and the response was along the lines of "AI is our future and it saves me time" (of course in exchange, it costs me a lot of time!)

Answer 1

Create a common response so that you can easily reply to their emails:

Hi [So and So]

I do not appreciate receiving emails that contain AI generated content. It wastes my time, and makes me feel like my work and time are not appreciated. If you would send the prompt that you gave to the AI instead of the AI's response to your prompt, I would appreciate it.

Moreover, I have found that AI generated emails often contain contradictory/wrong information. Because I am unable to tell which parts of this AI generated email are real and which are AI hallucinations I cannot answer it as generated.

Best of luck, Cloud

And this is the hard part. This, and only this, is the response that you send to anyone who sends you AI generated garbage.

Additionally, you should go to your boss and write a formal complaint about this.

Answer 2

I'm going to recommend something that has been a pet peeve of mine, because I think it might actually be a good communication strategy here.

Once you see that this is the kind of request you received, skim it quickly to identify one question that looks relevant. Respond with a short answer to that. Just send it, and let them follow up with additional questions if they actually have them. Put the ball back in their court. Don't mention what you didn't understand, unless you think it's critical to the task.

When people respond in this way to a well thought out email with multiple important questions in it, it's frustrating. But one of your problems here is that you don't know what the important questions are, hidden behind the "wall of text". Make a quick guess, respond with something that might be helpful, and let them do the work of highlighting what's important to them, if they wish.

This might not stop them from sending these sorts of messages, but it might help you work together.

Answer 3

Here are several options that I have used when faced with lots of undesirable work being dumped on me:

Let your boss decide

At the end of the day your boss needs to be aware of how you are spending your work day. If you are spending/wasting let's say 50% of it processing random slop AI requests that are not contributing to the bottom line, then they should be informed.

Your boss can in turn raise this problem to the bosses in the other departments that are responsible for this increased work load and from there leadership will hopefully make a decision on who should be handling what.

Pit requestors against other requestors

Allow only a small numbers of hours per day to handle AI slop requests, let's say 2 hours for the sake of this example. It becomes first come first serve, and once you hit that limit the rest have to wait till the following workday.

This approach works best when paired with some kind of ticket system that is public. Each request gets a work ticket with who requested it and the entire AI slop is attached to the ticket.

Then as people complain you direct them to the long list of other requests and their requestors who are taking up all your precious time. The goal being to get the requestors to fight with each other claiming their work is higher priority until some kind of arrangement comes down that limits the amount of AI slop they can dump on you.

Ignore the requests until they follow up

Create a Follow-Up folder in Outlook and take all those random requests as they came in and toss them in there and ignore them. For some (hopefully small) percentage of them people will send a follow up email requesting status on a previous request. At that point work the request. It is amazing how many requests that there will be no follow up on since the request never had any value to begin with.