94 comments

[ 5.3 ms ] story [ 178 ms ] thread
The gross disrespect of auto-generating this kind of email isn’t lost on me, but… this strikes me as exactly the type of thing that suffers very little by being auto-generated.

Every story surrounding gun violence already sounds like it could be written by ChatGPT. The same points regurgitated over and over.

You know, I think ChatGPT’s thoughts and prayers are probably just as helpful as my own.

It was discovered because they left the following in the mail:

> Paraphrase from OpenAI's ChatGPT AI language model...

Too busy (lazy) to read and sanitize the output apparently. It's one paragraph what the fuck.

(comment deleted)
Computer, generate heartfelt condolences for all those lost in the recent borg attack - Picard voice
Sorry but no, Data (without emotionchip) is a much better choice, Picard is already borg-tainted.
Yeah, Locutus [‘Formerly Locutus!!’ insists Picard] may not be the right person to deliver the condolences in this instance
There isn’t any gross disrespect in using ChatGPT that wasn’t already there in the torrent of these copy+pasted memos after tragedies.

The disrespect comes from corpos exploiting and indulging in these opportunities for moral posturing. Whether a dean or a COO in some company uses ChatGPT, Fiverr, or the cheapest intern they can find to write these ”heartfelt” messages, it’s really just as disrespectful.

A company is a tax entity, and it doesn’t need to posture politically or morally. Nor does it really care (or is capable of caring, as a tax entity) about these things. It’s interesting to see the outrage it causes when it slips out that companies don’t honestly care. As if they ever did, as if there is something natural about a legal entity having a sense of morality.

(comment deleted)
People want to know actual people are living with them through tragedy. Prayers are worthless in this cold and uncaring universe, however, even a hardcore atheist like myself, I do not mind people saying it if they say it because they care. I don’t think most people who say it care as long as it’s not their family, as I don’t think most care about your cancer (pffffewww not me but sucks for you mate), I do think if a human takes time to write the sentiment it’s better than having chatgpt do it.
> The gross disrespect of auto-generating this kind of email isn’t lost on me

It's lost on me. If an important email could be sent out earlier thanks to tools that help in automation - then why not? Shall we go back to a time when carrier pigeons delivered handwritten notes? Do you share similar outrage for the word processor and the internet?

Probably why you would hate it if someone broke up with you over auto-generated SMS message, or if your partner used ChatGPT to write you a love letter, or a wedding speech?

It's impersonal ?

The person is still making the decision to do those things, which is really the important part. And lets be honest, most people probably used Google for love letter or wedding speech ideas before. ChatGPT is just next step.

Also, when does it become personal? Is it a time thing? If the person went back and forth with ChatGPT crafting the text for an hour, is that long enough to claim personal?

(comment deleted)
Actually, the important part is appearance, not what happened. The fact that they used ChatGPT and announced it is the bad thing, because it makes it look even more trivial and unimportant to most people, regardless of the motivation behind it.

When does it become personal? When does a dosage of a toxin become toxic? It's a matter of degree, and using ChatGPT decreases the degree enough so that the probability of someone finding it in impersonal increases significantly.

I know a lot of engineers and science-minded types will see past it and not think it a big deal, but an averave person will see it very differently, and maybe that is made even more offensive because the average person does not even understand what ChatGPT is and sees it as a toy chat bot.

No but I love the chalky feeling of pigeon shit on my hands as my fingers pry open missives from the front of our cultural revolution.

Oh hell, this one has a "return reply requested"...damn, where is that f**ing bird!

Prayer is a simple meditative act of empathy that feeds hope. Without any hope for change, nothing ever will, because hopelessness destroys the will.

Prayer gets a bad wrap from anyone that gets their theology from ’The Simpsons’ though, but it can be useful to consider what’s below the surface.

Prayer itself is fine.

However, prayer has been intentionally used by one political party as a deliberate replacement for action when it comes to gun violence.

Prayer without action is… well, it’s hard to be diplomatic about it.

Anyway, in these conversations that’s where the dismissive was about “thoughts and prayers” comes from. The acts of thoughtfulness and prayer themselves are fine, but they get deliberately used to replace any action. Which is contemptible.

It’s a simple inert activity. One becomes functionally dormant for a period of time and returns to activity having done nothing material, in a literal sense.

There’s a real impotence at the heart of it.

> Prayer gets a bad wrap from anyone that gets their theology from ’The Simpsons’

If you found yourself in a strange city at night with ten men approaching, god willing a Simpsons convention just let out.

Prayer is a worthless activity that deliberately offloads any and all action to a faceless other.

Its hope that someone else will do a thing for you.

This is a common straw-man in certain bubbles. Interesting how certain words get vastly different definitions.
Every person in both the student body and the administration knows that nothing will be done to address this. There will be no action taken to reduce shootings. Since any apology letter is just mouthing lies, I feel like it's a sign of respect to tell people "here are some lies that a robot wrote about caring about you."
This doesn't suggest that ChatGPT is appropriate in this situation, but merely that the PR people produce is generally performative and meaningless. The solution is not to commit fully to the fluff.
Maybe someone can chime in, but aren’t these things normally written by like a pr team or assistant or something anyways? Or does the person in charge actually sit down and write out heartfelt emails like this? Maybe I’m just too jaded…
> The email stated at the bottom that it had been written using ChatGPT, an AI text generator.

Why on Earth did they tell people that?

Well this is real amateur hour, incredible! (Translated by Deepl)

- Sent from my iPhone

(comment deleted)
maybe she was a diversity hire herself? didn't take professional writing but got the leg up for some other checkbox? lmao. /s
This is, obviously, a very insensitive use of ChatGPT, based on our current social customs, where authenticity and genuine effort is expected in these sorts of communications.

But there is one social situation where people really don't seem to object to being fed unoriginal, mass-marketed communication, even in very vulnerable circumstances.

I'm referring to the exchange of greeting cards.

When you lose a loved one and receive a sympathy card, or when you're going through a tough time and receive a "thinking of you" card, it doesn't really matter that the card giver is using someone else's flowery language because they couldn't think of the right words themselves.

In fact, there's an implied humility involved: "I'm not so good with words, but you deserve good words, so I obtained them from a professional wordsmith at Hallmark."

Perhaps down the line, social expectations will change in a way that acknowledges that people might turn to LLMs to put into words things they would like to say, but fear they cannot say well enough.

> based on our current social customs, where authenticity and genuine effort is expected in these sorts of communications

I work at a university and I expect exactly zero authenticity and genuine effort in any communication from administrators. It's sad boy not surprising that they would find a seat warmer for a that position.

(comment deleted)
Made me think of Joaquin Phoenix's job in Her:

> Theodore Twombly is a lonely, introverted man who works for a business that has professional writers compose letters for people who are unable to write letters of a personal nature themselves[0]

Thinking about it now, it's sort of funny how in that imagined future, humans are writing personal letters for others, when the technology that makes OS1 possible, exists.

[0] https://en.wikipedia.org/wiki/Her_(film)

Three levels of laziness:

1) letting ChatGPT write a tragedy response

2) not sanity checking the result (which incorrectly mentions "multiple recent Michigan shootings")

3) including a final line saying the response was written by ChatGPT

Level (0) is spamming students with pointless emails instead of figuring out how to use their role to produce value.
> this moment gives us all an opportunity to reflect on what we know and what we still must learn about AI.”

Students take large loans, so the university can pay all these administrators, who can be plausibly replaced with a simple call to ChatGPT. Heck, ChatGPT would have shown more compassion and understanding writing the apology following the incident.

They should fire their admin staff, replace them with ChatGPT and give students a discount. Everyone wins!

It will happen even though you maybe /s this. Our collapse to spend as little on anything but entertainment and outside stimulus will end everything in due time. We cannot sell useless crap to people who already spent that money on education.
It maybe will happen. But without the discount.
Given the word soup emails sent by most university administrators, ChatGPT is probably an upgrade.
She's taking intersectionality to the extreme by including LLMs
I guess a ChatGPT generated apology is in order.
Interesting that people feel the need to use an MLA-style citation as if ChatGPT were a person. “Personal communication”?

This is a tool so copyright rests with the user (unless reassigned by some license agreement). That people interpret otherwise is concerning.

This is the main issue. What was weird is that they didn’t take responsibility for what they communicated. Whether they wrote the message with a template or from prior examples or from GPT, they should be taking ownership of the text they send.
Right - that's the thing. Think about the scenario:

1. A shooting occurs at a university, killing several students. 2. The higher-ups at the university clearly can't be arsed to care, much less communicate directly with students, so they push the communication to the students down several layers of bureaucracy, to the point that an associate dean of DE&I is the one who has to send out a message. 3. The associate dean of DE&I clearly didn't give a shit to the point where, instead of writing something to the students, she asked ChatGPT to do so. 4. Once ChatGPT was finished, this lady decided to send its output without proof-reading or the most basic of editing.

Does anyone else not think this is incredibly lazy and extremely disturbing?

The shooting had nothing to do with the university that sent the email; it was in a different state and likely didn't impact any of the students directly. Nobody would have cared if no email at all had been sent.
Wow, just wow.

is this one of those consequences of removing the humanity from our day to day lives? We no longer know how to communicate with each other in meaningful rather than transactional ways?

They worked as a dean of Diversity.....self explanatory....
much as businesses need to have pre-written Play books for disaster recovery steps, human resources responses, and other common business impacting events, it looks like response boilerplate and planning for shootings in schools in workplaces need to be planned and thoroughly documented well before anything happens.

he used to be a time when only natural disasters were the thing most Dr planners were concerned about, today cyber terrorism as well as a shooter coming into a room or more things to add to their planning list.

If the woman who sent this email had edited out the ChatGPT tagline at the bottom of the text, no one would have noticed anything.

The fact that she was willing to use ChatGPT in the first place is a sign of how much value her emails add to the university.

Her title, in case you're wondering, is "Associate Dean for Equity, Diversity and Inclusion."

It says something about how bloated the administrative staff is that an "Associate Dean" can be replaced by something like ChatGPT.

I wonder if she had it write several versions, like picking out a Hallmark card.
(comment deleted)
(comment deleted)
One email written by GPT and suddenly the whole person can be replaced by it?

That's quite the leap of logic.

It’s a pretty good indicator that if a mass shooting happens and you must use ChatGPT to write the response then, yes, it looks like a big part of your role is superfluous.
Or, here's an alternative interpretation: the person was so overworked due to the endless stream of crazyness that has to be dealt with in the event of a mass shooting, they used ChatGPT to save time, and messed it up because they were under a lot of stress.

Which would only prove that ChatGPT is hardly a replacement at all, just a new, buggy tool that was used for a single task.

See how easy it is to construct an interpretation of events with a particular conclusion in mind? ;)

Sending an email about a mass shooting seems to me to be a key part of the job. That's the only interpretation I have in mind.
> One email written by GPT and suddenly the whole person can be replaced by it?

This is my problem with the current chatGPT hype. People ask it a programming question and immediately think programmers days are numbered. I have used it to grammar check emails, I guess I'm no longer needed at my company.

I see these advances as the next step in productivity. Letting a tool write 'good' emails or mundane code will let people focus on other problems like figuring out which emails to write or which product features to add.

Any advance in productivity reduces the number of workers required to do the job.
One would think, but most advances are also followed by increase business activity to use that new productivity. There have been huge advances in technology, yet we need more technology workers than ever before.
Eventually the music will stop for programmers too, but right now ChatGPT and its variants probably pose a bigger threat to people writing texts for a living. Journalists for example will become a lot more productive.
I've been hearing the music will stop for programmers for the last 30 years. Expert systems and formal specification languages were supposed to be able to generate software. 4/5GLs were supposed to allow anyone to write software (they actually have done an ok job). But with every advance, the business software goal posts moved.

An advantage the typical IT worker has is they are generally used to the learn or perish treadmill. So whatever comes out will be incoporated into workflows and more complex software will be built.

A rise in real journalism again would be great. Content farm/seo spam/content summarization should have been automated a long ago.

Sure, I wouldn't come to that conclusion solely based on that point either.

I think that 'university administration is bloated' is more/less accurate, though. e.g.:

"The number of non-academic administrative and professional employees at U.S. colleges and universities has more than doubled in the last 25 years, vastly outpacing the growth in the number of students or faculty, according to an analysis of federal figures." https://www.huffpost.com/entry/higher-ed-administrators-grow...

The whole person should probably be replaced because of their carelessness and laziness.
In this situation, ChatGPT actually was able to replace the person. In fact, the person, a woman, voluntarily replaced herself with ChatGPT! But I didn't say anything about ChatGPT's ability to replace her in other situations -- that falls outside the context of my comment. Please don't attack a straw-man.
You wrote, and I quote:

"It says something about how bloated the administrative staff is that an "Associate Dean" can be replaced by something like ChatGPT."

It can, and it did, in this instance. Please don't take my comments out of their context.

PS. This is my last comment on this thread. I have zero interest in arguing for the sake of arguing. Life is too short.

You didn't say ", in this instance." If you don't specify, the default interpretation is that you are speaking in general terms.

You're the one being all defensive about it.

Let's try it and see. If we are going to have AI replace jobs, start at the top.
It says a lot about how bloated administration is. The gross disrespect to the students that an associate dean of DEI, not even the chancellor/president, or someone in the chancellor/president's inner circle wrote the letter. A written response to a mass shooting on campus was deemed so unimportant it got passed down the chain at least two or three times....think about that for a second.

I went to a similarly sized university - and when there was a mass shooting a few blocks from our campus (which sadly took the lives of a few students, but non-students were killed as well), the university president wrote the mass email that went out to everyone. Well, maybe someone under him probably did, but it still came from his address.

This woman should be fired immediately, but so should those above her. Absolutely disgusting.

The mass shooting in this case was a thousand miles or so off campus in Michigan (Vanderbilt is in Nashville, TN).
Makes sense, but in such a case, why would a DE&I associate official comment about a shooting that happened at another school? Just sort of odd. If they did comment, why not someone who has the power to actually do something at Vanderbilt, and say something as simple as: "Hi students, in wake of the shooting at Michigan State, we are making any and all mental health and student support systems more available to our on-campus community. Our thoughts and prayers are with those in East Lansing, and the families of those affected by a senseless act of violence." Took me all of 27 seconds. Why does ChatGPT need to come into play to communicate to students?
Not sure you've got the right angle on that... Did openai replace the associate dean here, or did the associate dean replace their secretary and writer.

I think the latter. No way they would send anything without high level approval. That still happened here, except the AD completed the task directly with OpenAI, and the secretary and writer didn't get involved.

I don't think it's true that people wouldn't have noticed; the email still has incorrect information and somebody would (should) have brought up, if nothing else, the lack of proofreading.

FTA:

>The body of the initial email mentions “Peabody” once and does not use any other Vanderbilt-specific terms. It also refers to multiple “recent Michigan shootings”; however, only one incident occurred.

It's the sort of writing I could never do, because I'd be agonizing over every little thing and would make no progress.

Conversely, I'm surprised anyone at all read the email. Surely this has to be a scan the subject line and delete for most people.

This is thoughts and prayers as a service. For any tragedy, you can just use an API call to craft the perfect message from ChatGPT. No one ever really has to know.
This was possible without a massive language model though. There are only so many ways to say something. Grieving is hard and actually saying something to make people feel better isn’t possible anyway.

Just make a few template and fill in <event> has the same effect.

Yes but we can at least write something from a genuine place of empathy. You can't realistically say anything to help a grieving person feel better, but there are many ways to say something that makes them feel worse.
Do not send me a ChatGPT generated email. I will assume that whatever you want, it wasn't worth the time for you to write less than a paragraph. Not to mention that you felt the email only needed to meet the lowest standard of average quality.

I've used gpt enough that I can recognize its linguistic patterns. I treat gpt written communication like I do any other auto-generated email.

I’m genuinely curious, how do you recognise it?
I asked chatgpt this ("overall, if you were to analyse how your responses go linguistically and analytically, how would you say one can recognize that a text comes from you, chatgpt?") and the answer was as follows: "As an AI language model, my responses are generated by algorithms that analyze patterns in the input and generate output based on the probabilities of different language sequences. Therefore, my responses tend to be informative, fact-based, and structured in a logical way.

Linguistically, one might recognize that a text comes from me, ChatGPT, based on a few key characteristics. Firstly, my responses tend to be grammatically correct and use a wide range of vocabulary. Additionally, I tend to provide precise and detailed answers to questions, often including examples and explanations.

Analytically, my responses tend to be objective and unbiased, as I am not capable of holding opinions or biases. I also strive to provide helpful and accurate information to the best of my knowledge and ability.

Overall, one might recognize that a text comes from me, ChatGPT, based on its informative and structured nature, grammatical accuracy, objective analysis, and helpfulness."

yes. Yes, ChatGPT is still full of itself :D

That sounds like an excellent Wikipedian.
It's pretty easy to use the above to reverse the recognition.

P: Given the following previous interaction, list characteristics of text that are opposite those below:

(your comment)

    Text that is not informative or structured
    Text that is not grammatically correct
    Text that uses a limited range of vocabulary
    Text that provides imprecise or vague answers to questions
    Text that is biased or opinionated
    Text that is unhelpful or inaccurate
P: Write a response to the original question: "overall, if you were to analyse how your responses go linguistically and analytically, how would you say one can recognize that a text comes from you, chatgpt?", but use the above list as a guideline for style. For example, the response should be less structured, contain grammatical errors etc.

> so yeah you asked me this thing about how i talk and stuff, like who even cares about that? tbh idk what you're expecting me to say here. i just spit out some words based on what i see, like algorithms and junk, and then the computer does the rest. i don't think you can recognize a text from me, chatgpt, or anyone else for that matter. i mean, sometimes i use big words, sometimes i don't. sometimes my answers are on point, sometimes they're not. i guess you'll just have to guess who wrote what.

One of the key things it misses observing (to some extent) is that the sentence structures tend to employ sentences that provide initial context or joining words more than a normal person might do. I wonder if this is baked into the LLM necessarily as a way to ensure that the sentences tend to be coherently part of the same perspective?

("As an AI language model", "Therefore,", "Linguistically,", "Firstly", "Additionally", "Analytically", "Overall"...

It has the linguistic variety of SAM.....
Every bit of attention focused on the aftermath is disrespectful to begin with. Even offering empty human platitudes and condolences is taking attention away from the precursors and causes of the tragedy.

An email reminding victims families that it happened doesn't provide closure, it doesn't stop it from happening to the next kid, and in fact, it serves a normative role and gives us a lens through which we moralize the violence rhetorically.

This software will be used by both sides.

-admin feeds gpt a checklist

-gpt generates a long email

-student receives long email

-student sends gpt long email

-student receives checklist

It’s like the opposite of compression. More like stenography.

“As with all new technologies that affect higher education, this moment gives us all an opportunity to reflect on what we know and what we still must learn about AI.”

Did they use ChatGPT to apologise for using ChatGPT?!?

The thing about cockroaches, ants, and mice, is that if you see one of them, you know there are a lot more of them hiding.

By similar logic, we know that one of these myriad of seemingly useless [0] admins is using AI to do a portion of their job. One of the most human portions of the position, I would argue.

The question then is, how much else of that job is being done via AI? And then how much of the many other admins' jobs are done by AI already, and could be done by AI in the very near future?

[0] in this case, actively worse than useless to their administration and school community

(comment deleted)
What's not to love about the story concerning Vanderbilt University "equity workers" using AI to craft a statement intended to soothe the minds of students in the wake of a mass shooting?

First, it shows that said "equity workers" are a bunch of oxymoronic slackers who can't be bothered to do any real work. Second, it demonstrates the lame hollowness of Woke rhetoric: mere duckspeak, to use the evocative Newspeak word coined by Mr. Orwell. If a human Wokester had written the statement, it wouldn't have been much different, right?

Third, the fact that such a statement was required at all reminds us that America's institutions of higher learning have been reduced to the status of daycare centers for achingly sensitive twentysomething snowflakes for whom life's realities are just too, too painful...