I keep hearing stories and podcasts about how people are using generative AI at work - how it’s “so efficient” and they’re having more time for other tasks, etc. Whenever I’ve tried it for various tasks I haven’t found it helpful, though.
Example. A (young) coworker uses it to write the first draft of emails, but that’s a task I’m very fast at, or it’s a task where I need to impart knowledge that is in my head and the AI wouldn’t have that anyway.
I think a few people used it to do their performance write ups, but I don't think it would work for our actual work products.
Someone did say they were testing AI on certain low level documents, but because of my job someone with a specific authority has to sign the end product. So it would only automate a certain percentage of work and a human is still the end.
Post by turkletsmom on Jun 13, 2023 9:18:51 GMT -5
I haven't but my sister, who is a middle school assistant principal, was just telling me how she tried it for the first time to write letters of recommendation for students that were transferring to private schools, etc. She said it worked great and she barely had to do any tweaks.
We are developing an enterprise strategy. Generative AI is going to transform work, and we have to really figure out what it means across employee and customer life cycles.
but we had a legal seminar on unrelated topics and there were major red flags raised about privacy and discoverable documents related to use of Chat GPT. the lawyer presented case studies that are in process about unexpected ways companies are getting themselves in trouble because of lack of privacy in chatgpt specifically.
My favorite email all school year came from DD’s AP World Studies teacher. She sent a group email to all her students and their adult. She told the parent/adult not to waste her time like the students have. If they want to have a serious talk about plagiarism and chatGPT they can contact her. She failed 1/3 of the class for using it. She has sent a million and one text, emails, and notes telling families to not use chatGPT.
Post by RoxMonster on Jun 13, 2023 10:14:26 GMT -5
No, my company prohibits it due to it being open source and security/privacy all that. They are looking into creating our company’s own version of ChatGPT though.
but we had a legal seminar on unrelated topics and there were major red flags raised about privacy and discoverable documents related to use of Chat GPT. the lawyer presented case studies that are in process about unexpected ways companies are getting themselves in trouble because of lack of privacy in chatgpt specifically.
Yeah. I’ve avoided it because of this. We have very strict privacy rules (not just people privacy). So I’m scared to put something in least it clue someone in that it’s something I’m looking at.
which is likely visible from my Google search history. But I’m very gun shy.
I used it to generate a first draft of a year-end message. I say basically the same thing every year, and I was running out of new ways to say it.
Honestly, it was awesome. I had to edit in specifics about our accomplishments over the past year, but all the fluffy "we have had the privilege of working with great people, thank you for your dedication over the past year, blah blah blah" was taken care of for me. It's much easier for me to edit than to generate from scratch. It gets rid of the daunting blank page, even if I end up heavily editing the entire first draft.
I work in scientific publishing and it has been a HUGE topic of conversation in terms of if, and how, AI can be used ethically in scientific writing. Ultimately it can't generate research on its own, and we've mostly landed on the authors being fully responsible for the final product. If they want to use AI to help with their introduction or something it's fine as long as they assume responsibility for any potential plagiarism, and it is on them to edit for inaccuracies. One journal specifically requires disclosure, but the other does not.
My boss actually just gave a presentation on how to use it in marketing. I have used it with mixed results. Several social media schedulers have AI elements, and I generally find those suggested posts cringy or not super useful. I've used both Bard and Chat GPT to see what the longer copywriting capabilities are. So far the most useful thing I've done with it is plan some vacation itineraries for personal use.
ETA: It isn't as straight up AI as I think you are thinking of it but I would say that I use Grammarly's advanced editing - which recasts sentences and does some writing - daily. I am the human and make the decision on what is used but that service is the equivalent of a FTE on my team. Big fan.
Well, I'm an academic librarian and we are trying to figure out how to teach / address AI with students. Outside of ChatGPT, there are some AI applications that are very good at finding scholarly sources, but I still need to teach students how to do a proper search using our databases.
ChatGPT is still making up sources, which we are then asked to help find. Such a waste of time. Now I ask students (and faculty!) where they found the sources before doing any work.
I've used it a few times with success. If you copy in the transcript of a Zoom meeting it will spit out a 2-3 paragraph overview of the discussion. That's great for notes. I used it to create a draft outline for a 3 workshop series on Authentic assessment that matched almost exactly the one my colleague drafted herself I've also used it a couple of times to write haikus about projects just for fun, such as ...
“With sorrow—for this Court, but more, for the many millions of American women who have today lost a fundamental constitutional protection—we dissent,”
Post by mustardseed2007 on Jun 13, 2023 10:38:45 GMT -5
There's a viral case right now of a lawyer using ChatGPT for legal research to find cases. Allegedly ChatGPT found fictitious cases for him. When the judge and opposing counsel couldn't find the cases, he allegedly went back to Chat GPT to find them and ChatGPT allegedly wrote content for the fictitious cases. Lawyer twitter is following the sanctions hearings with bated breath.
I use it to generate ideas for catchy article titles and other simple things (where the information is public/non confidential) if I'm stuck. I don't use it on my work computer or for anything more complicated because we deal with a lot of confidential information in my organization.
I want to echo toepick – if you're using AI for any type of research, make sure to check the sources. I've played around with Bard and ChatGPT. They both invent books, articles, primary source quotes, etc. And they make up statistics and data if they can't find it. So just be careful to double check info!
Post by definitelyO on Jun 13, 2023 11:04:16 GMT -5
no I don't - I haven't taken the time to dig into it. there are other areas of our business that are dabbling in it. helping to write new job descriptions, some marketing, etc...
Post by mrsslocombe on Jun 13, 2023 11:15:23 GMT -5
There currently isn't much use for it in tv editing thank god, though there have been some scarily decent short videos done in testing for commercials.
I did realize the perfect use for my husband-drafting the bullshit yearly goals he has to submit to HR.
I haven't but my sister, who is a middle school assistant principal, was just telling me how she tried it for the first time to write letters of recommendation for students that were transferring to private schools, etc. She said it worked great and she barely had to do any tweaks.
Something about this makes me uncomfortable, and maybe it’s the fact that LORs require personal information about students and it’s not secure? I don’t know, but I would tread lightly with this.
I’m a teacher and I often write my own texts (since I don’t have a text book and I need different reading levels). ChatGPT has been a godsend for this. I can type in “write a one page text describing the causes of the Haitian revolution on a 7th grade reading level” and get a useable text that I can then edit as needed. I could do it, but it’s much faster this way.
"Hello babies. Welcome to Earth. It's hot in the summer and cold in the winter. It's round and wet and crowded. On the outside, babies, you've got a hundred years here. There's only one rule that I know of, babies-"God damn it, you've got to be kind.”
I haven't but my sister, who is a middle school assistant principal, was just telling me how she tried it for the first time to write letters of recommendation for students that were transferring to private schools, etc. She said it worked great and she barely had to do any tweaks.
I don’t like this idea. One flaw of AI is that it’s very generic — it writes a very bland recommendation that doesn’t actually capture a kid adequately. I’d be upset if someone used it to write a rec for my kid.
"Hello babies. Welcome to Earth. It's hot in the summer and cold in the winter. It's round and wet and crowded. On the outside, babies, you've got a hundred years here. There's only one rule that I know of, babies-"God damn it, you've got to be kind.”
I’m a teacher and I often write my own texts (since I don’t have a text book and I need different reading levels). ChatGPT has been a godsend for this. I can type in “write a one page text describing the causes of the Haitian revolution on a 7th grade reading level” and get a useable text that I can then edit as needed. I could do it, but it’s much faster this way.
Yes, I recommended that our computer science teacher use it when she was trying to get students to generate short fables to use as templates for a Mad Libs programming project. Students were getting stuck on writing the story, which wasn't the standard they were being assessed on at all. Chat GPT allowed them to create stories that met the assignment specifications, about topics the students chose, and from there they could apply the actual CS lesson.
That said, most of my work with Generative AI has been teaching both students and teachers what it can and can't do. I've spent more than a few hours already this summer compiling a list of links to cautionary tale stories that will probably make their way into some of my fall research skills lessons. I'm glad I don't have to touch the plagiarism piece with a 10 foot pole, and don't envy the educators here who are navigating that piece.
I haven't but my sister, who is a middle school assistant principal, was just telling me how she tried it for the first time to write letters of recommendation for students that were transferring to private schools, etc. She said it worked great and she barely had to do any tweaks.
I tried it out to do exactly this task. It worked pretty well (I didn't use the GPT product, I just wanted to see what it could do).
I find these types of letters incredibly hard to write, not because I don't want to endorse the kid (or adult) or because I don't know them, I just have a hard time capturing it all in the best way. And I'm a good writer(!)--recs (and cover letters!) just aren't my strength.
I haven't but my sister, who is a middle school assistant principal, was just telling me how she tried it for the first time to write letters of recommendation for students that were transferring to private schools, etc. She said it worked great and she barely had to do any tweaks.
I don’t like this idea. One flaw of AI is that it’s very generic — it writes a very bland recommendation that doesn’t actually capture a kid adequately. I’d be upset if someone used it to write a rec for my kid.
When I tried it, I was able to put specifics about the kid into the content field and the generator included all of it. That said, some of the wording was a little wonky; I was asking it to write a rec for a student that had been in my honor society who was applying for a specialty extracurricular. It wrote the letter in such a way that it seemed I was recommending her for a specialty school.
No. Unless it can tell me why an insurance company didn't pay a clean claim, I don't have much use for it in my office at the moment.
On the flip side, I'm reading that insurance companies are using it to deny even more stuff than they already do.
I have colleagues who it to write their appeal letters. Works decently well and when they paste in the chart notes, the letters are very detailed. I don’t use it for that purpose because I’m not comfortable with the privacy situation, but it is becoming more common.
I SAHM and use it for all sorts of things. Most recently to help me write a letter for funding for a non profit. It’s amazing. I read what ChatGPT writes and use what is relevant or it gives me ideas on things I had not thought of.
My initial instinct was to refuse it but it’s not going away. We need to learn how to use it ethically and effectively.
My H used it to draft a letter to the IRS on behalf of a client - I think it was to argue something that came up in an audit. He was amazed at how well it did and said this would significantly reduce the time it takes going forward. These letters use a lot of legalese speak and code citations and can be tedious to draft.
Obviously it still needs to be read through and edited, but it is super helpful for pulling in the citations, I can see a lot of applications in technical writing.