Hacker News Reader

Comment by felipemesquita | Hacker News Reader

If you’re just wondering how students knew that the email was written using ChatGPT, the last line of the email weirdly reads “Paraphrase from OpenAI's ChatGPT AI language model, personal communication, February 15, 2023”.

View on HN

That's the only thing they did wrong. Why mention you used ChatGPT? It's like emails that end with "sent from my iPhone" like who cares what technology you used to send an email?

With “sent from my iPhone” it’s an excuse for brevity and/or typos. Consider that the iPhone’s keyboard was a downgrade from the Blackberry that was previously standard among business users. The automatically-appended disclaimer saves the sender a bit of stress.

that's one take.

Personally I think it was motivated by trying to create a marketed feeling of exclusivity , drive a stake between Those With and Those Without, and alienate those without into buying an iProduct and joining The Club.

Maybe a little bit less sinister than that. Just a way to promote the product in every email originating from their devices

Regardless of the reason, I think less of anyone who doesn't turn it off. Made me cringe so badly when I sent an email from a new email client that added something similar and I hadn't checked the signature.

One reason I feel that way is that the only time I'd see it in a work context is if it's a phishing email.

I see it all the time in a work context. At three different jobs.

No one really cares where I’m at. I have my laptop signature set to:

    Thanks,
    Kayodé
    
    $NAME
    $RANK
    $SERIALNUMBER

I leave it there even if using “Thanks” isn’t entirely correct and my name is repeated below it. Everyone else here does something similar.

My phone signature is the iPhone default. Just like everyone else.

The Blackberry had “Sent via Blackberry” before iPhones exsisted.

Or more simply to convey "I sent this email while on the go so without access to my pc/office/documents/files/etc". Gives important context.

Even though I could change it, I never bothered to remove the default signature. It works pretty well as a disclaimer because most people don’t bother to remove it. Therefore, you’re able to signal you’re not replying in an optimal manner without the other person feeling slighted.

“Hey, I’m sending this email from my phone” is pretty much a necessity when dealing with clients and such.

I’m grateful the default Mail sig works as a graceful catch-all rather than fitting in an awkward mea culpa in the body or (worse) slapping in some boilerplate sig that might as well read “this email is unreliable and so am I” to certain types of people.

My theory is that the person responsible for publishing the email disagreed with using the chatbot, and included the attribution statement in a form of malicious compliance.

For clarity, as multiple replies seem to take this as a proofreading mistake: ChatGPT doesn't output such a phrase unless explicitly asked to.

They're just citing their source LOL!

Maybe they thought it would be worse if someone found out and it was exposed that they didn't cite their source.

It's not a source in this case - if I understand correctly, they used ChatGPT to rewrite a message written by them.

I understood it differently. Saying the email is a “paraphrase from ChatGPT” seems more like they had ChatGPT generate it, and then modified or “paraphrased” it themselves

Models that recognize chatgpt also seem to pick it up easily though [0]. Adding it in the email is quite weird though. Or maybe they agreed that is was fine to do so (generating this email by an AI), which is indeed tonedeaf.

[0] https://filteroutai.com/validate/114ddca7445e6b5119715111623...

This is the mind boggling part. What were they thinking? Was this copy pasta from OpenAI and didn't proof read it?

University administration is a kafkaesque environment where enthusiastic compliance with a ruleset is of foremost importance. In this context, citing a source is covering your ass. If anyone has a problem with the wording, it’s GPT’s fault. It probably never occurred to them that citing the source could get you into trouble.

ChatGPT doesn’t insert itself as a citation, it was added deliberately. For what reason is the question…I don’t say I use spellcheck when I write something with Word.

Yes, nobody would know. They weren't trying to hide it.

To their credit, they attributed the assistance by the chatbot. (I don't think they carelessly copy pasted and did not notice the chatGPT notice).

Of course, humans expect honest and sincere communication in times of grief and not automated responses, so that's on the Deans for making poor decisions (just as poor as it would be to hire a PR firm to compose something for them --it probably happens but just does not get announced)

Reply on news.ycombinator.com