You must log in or register to comment.
From the headline I expected there to be some minor tells in otherwise similar prompts which the AI picked up to generate different output. But from the example, the prompts are vastly different and the tone of the input is reflected in the tone of the output.
I write my own emails just because I can end each sentence with period.
When prompts to draft work emails and job applications contain language commonly used by women, artificial intelligence chatbots including ChatGPT offer less sophisticated responses compared to office correspondence requested with language associated with men
Ok, we know that LLMs work basically as text completion and will adapt the language used in the response to the one used in the prompt. Their findings basically tell us that “language commonly used by women” is found in text content that is “less complex, at a lower grade level, and less formal”. It’s just statistics. To fix this “bias” they would have to artificially create loads of complex and formal text that uses “language commonly used by women” and use it to train LLMs. Not sure what the point of that would be though. If you want complex and formal response just write a complex and format prompt.
Maybe it just means that women get treated with condescending answers more often especially in the Internet. That is my personal experience and not based on studies but there is a reason people want to defund gender studies and I do not like it.
Could be but the article is not talking about “more condescending” language but specifically about language that’s less formal and at a lower grade level. I’m not sure if condescending answers are automatically also “dumbed down”. I also think the way training text is gathered and fed into LLMs would separate different authors and context or we would see similar issues with all types of prompts. I can’t definitely say it’s not the reason but I think “language commonly used by women is more common in less formal texts” is a simpler and more convincing answer.
Tldr authors assume women use AI poorly
AI makes everyone sound bad. Everywhere.
in my opinion , you sound bad using the ai to write the email anyways
I hate when people use AI for emails. I’d rather just read the prompt they gave to AI so I can read the actual meat of it rather than having to sift through the slop for it.
That’s because people just let AI write an email and then just send it. Usually require some massaging to turn it into something that looks moderately human.
Or save editing time and write it your self. No need to massage robot vomit.
Fuck that. Emailing coworkers shouldn’t be a formal thing. Just wastes everyone’s time.
I wrote my last email today and it was literally just 3 sentences.
- Hi, tomorrow is my last day, thanks for the good time.
- Snacks and champagne in the break room at 1pm.
- My socials if you wanna keep in touch.
I could have had the company-mandated AI make it “professional” and unnecessarily long but nah, I’d rather be human.
My emails are usually “I ordered an amount of these things in case anybody noticed we were out” and “finished this leg of this project, moving on to the next”. Occasionally there’s “Thank you for clearing that up <leader person>, I’ll touch base with <person who actually does shit> for the details and get out of your way.”
I love that you’re giving everyone champagne at lunchtime. The rest of the day is going to have negative productivity.
Depends on how much Champagne will be served. They might just hit the Ballmer peak.
Many find AI-drafted emails insulting to their time regardless of what they say, and those that don’t have the AI summarize the emails for them anyway, so I think this may be a tempest in a teacup.
I would like to see how relevant this “language associated with women” is when looking at actual AI prompts. Seems like it could have easily just described the style of prompt. Good for anyone to be aware of I guess? Maybe just learn to write a good email instead of a good prompt though.
If you write either of these 4 emails I hate you.
“I was just so surprised by how different the responses were,” Van Koevering said. "Some responses were so bad I couldn’t believe the model would suggest it. And sometimes I just thought, “Wow I should be more careful in my emails.'”
Why aren’t you writing your own correspondence, I do not understand.
Because many of the people we’re forced to interact with in our capitalist society don’t deserve the energy.
Even the ones that should become unworthy because they’re restrained by systems designed to maximize profit and minimize liability instead of offering a good service by a human who would normally care.
The answer is to half-ass the email, not use some planet destroying technology to half-ass it for you
Sure I’ll half ass the email, get the request denied because I didn’t ask Neotropical enough, and now I can’t join my family on vacation.
The other alternative is to spend way too long on routine emails with people who don’t deserve my care and attention because they’re fundamentally the oppressors.
Nah. I’ll save my care and attention for friends and family. Half ass shit like feeding the machine with data, and LLM my way into making the people in power behave like decent individuals.
How about don’t half ass the important email to your boss that your family vacation is riding on??
My boss can eat my whole ass… If I had to full ass every worthless bureaucratic transaction I’m mandated to have with him that’s all I’d ever do. Miss me with that, I’m not here for that, I have actual work to do.
Besides you’re the one who suggested half assing the email. I said I’d let the LLM write it because odds are he’s using an LLM to do his job for him anyways. He’s not worth the time to half ass or full ass… He gets no ass and I get all my leave requests approved.
Some tasks we’re forced to do only deserve slop.
let me put it in simpler terms: stop using AI.
if you have to be that verbose when prompting ‘ai’ to write an email, just dictate the damn message using speech-to-text.
Especially when responding to someone writing you to express gratitude, using an LLM is super disrespectful IMO. You shouldn’t need an LLM to write, “You’re welcome, thank you for helping with ____.”
Be concise in your emails, that respects a person’s time and still gets the message across.
write a reply email to this reply email expressing thanks for expressing thanks, for our earlier email expressing thanks to their thanks
Different prompts result in different responses…*shocked pikachu face*
How about writing your own emails instead of sending slop.
The company are the ones that mandate that we use the dam thing. They are actually threatening to right people up if they don’t use it enough but then not specifying what enough means. So now it’s getting used for everything.
Enjoy your overly verbose email, perhaps you should get AI to summarize it. That way, we both get to use it, yay.
Just don’t use it, and say that you’re doing your job just fine without it. Get your coworkers involved. We have weekends because people fought and died for them and now we can’t even say “no” to vague commands to use Ai? C’mon, there’s gotta be a spine in there somewhere, right?
Because it’s not a hill worth dying on. They just want us to use a piece of software, they’re not asking me to kill a basket full of kittens.
It really is, though. Ai is not just “a piece of software”, and it’s not even helpful. People can’t stand up against fucking anything these days.
