How do you use ChatGPT?

[foolsonyou]

I've been using ChatGPT to edit internal documents, proposals, meeting minutes etc for a while now.

Has anyone use it for academic writing? I've played around with sticking in half written abandoned papers but I honestly cant tell if what it spits out is actually good.

I'm not sure about the ethics (isnt it like a spell checker) and how detectable is it? If I use it to proof read is that going to set off AI detection at journals?

How do you use it at work?

Fuck AI. And fuck students who think they can pull one over on me by using AI when instructed not to, tripling my workload, and fuck any institutions and colleagues that facilitate it.

Also, if you don't know about the ethics of AI, get a clue.

If you can't pull that together yourself, why are you even in the job though?

I’m sure they can do, but ChatGPT can come up with a list of stuff quicker. So it’s a time saving thing. To come up with typed activities and scenarios as well as thinking of what topics within a session to cover might take a couple of hours if you did it yourself. AI will do it in a minute. With the amount of cutbacks in HE we’ve got to work smarter.

I could do a load of maths calculations, etc myself on a piece of paper….or I could use a calculator….

who pissed on your academic chips, @ColourlessGreenIdeasSleepFuriously ?

Wasting a bunch of fucking time marking essays where the students were explicitly instructed not to use fucking ChatGPT, and the normalisation of such brain-rotting tech that undermines the very purpose of what a university is for, is what has pissed on my chips

My ideas and writing are better than ChatGPTs! We’re meant to be producing original work, not mediocrity, no? Not re-scrambled stuff of other people’s (which is what LLMs do).

I'm a bit surprised that a pp was happy to feed qualitative data to Chatgpt for analysis. I'm fairly certain that's not ethical.

I personally lean towards the sceptical in relation to genAI and haven't used it in my work (though I have had a look to keep informed). We're explicitly told to use Co-pilot but I very much doubt students pay any attention to that.

Mediocrity scraped from my own highly original and award winning IP without authorisation or remuneration.

God, sorry to sound snitty, but if we don’t know how to do this without asking a machine then should we be doing this job?

I do appreciate that I’m in a field & institution where I have quite a lot of autonomy over what I teach (I’m in a humanities subject and a lot of my teaching is very research-led), but surely if we’re lecturing on a subject and at least a few years experienced in teaching, don’t we know these things without asking ChatGPT?

One of the nice things about writing a lecture is deciding what you want to tell the students that they can’t just get direct from a textbook. Maybe I’m just really lucky, but I like lectures precisely because it’s all me time: these kids have turned up and I’m going to tell them what I want to tell them for fifty minutes. And they’ll like it! 😆

I guess OP has not submitted any articles for peer review in a long time? For at least 2 years we’ve had “You must declare if you used AI” statements at submission, or asked to commit that AI not used to write the article at all, and finally in the last 12months we are getting notices as reviewers not to use AI at all to write the review (FINALLY).

I also use it to answer emails, but mostly when I'm intimidated by people senior to me and can easily spend 3 days editing an email that really doesnt matter.

FFS, If that is typical time management priority, no wonder so many university staff complain about working 12 hour days.

Adult DC uses chatGPT to make her emails more polite. I asked if she meant to deskill herself.

I sometimes tinker with using AI to do thematic summaries. It’s fairly bland in results but that can be ok when most open text data from surveys don’t get processed at all. I WANT to turn AI off when I do google searches but I haven’t found a tickbox to do that. Otherwise I would never use AI… I can figure out how to write a polite email. The process itself of trying to do something that gives me benefits. Why would I want to deskill myself??? And it's a stupid huge waste of resources. AI should only be used for things humans can't do as well.

I could do a load of maths calculations, etc myself on a piece of paper….or I could use a calculator….

If calculators made random shit up and hallucinated references and were wrong more than half the time you probably wouldn't use them

I had meta on whatsapp rewrite a letter for
me today and I’m so impressed, how does it do so instantly

You don't have to use google. I switched to ecosia.

We literally have an AI associate professor at our place giving sessions to staff telling us to use AI like this. We had a staff AI week recently and we all had to attend 3 sessions on how to incorporate AI into your work and teaching sessions.

I use it instead of google

Nope hate it hate everything about it. Am a happy dinosaur mucking about writing my own words on a field I helped to found and have worked in for 30+ years. Every time I am made to think about AI I go and look at my pension calculator (yes I know how bloody lucky I am)

It is vastly more wasteful of energy and water than google for results that are nonsense a non negligible % of the time. It relies on underpaid global majority labour to tag all the data including the grossest underbelly of the internet, with no labour rights or protection or mental health support. it reproduces all the big social biases against women, POC etc.

Bloody hell 😮 that really is dumbing down!

Surely one of the most important skills we have, that is also one of the most valuable to students, is the ability to synthesise lots of information and produce new and interesting overviews and big picture ideas? I often have students asking me after a lecture what book they can get to read up on what I’ve just been talking about, and I have to say well, nowhere quite as yet, because no-one else has yet synthesised all of this material in precisely this new way. They could go away and read a lot of the resources I’ve read; but even then what they produce out of it will be unique to them — that’s the beauty of original work.

That’s my key skill for them: to be able to have read those hundreds of books/articles and produce a new synthesis/research narrative out of that. I don’t think any AI has that capacity, at least in my field, because it’s that particular depth of what one has read and thought about for ten or fifteen or twenty years that makes it happen. I do appreciate that in other fields that might be different, though. I’ve heard of AI spotting new patterns in eg. medical research and data: it would be interesting to know from colleagues in the sciences/medicine if that actually does happen.

But in the humanities — nah.

I read one article about AI deskilling senior radiographers to junior level by making them less confident in their diagnoses

Did you ask ChatGPT to critique the article/research? 😆

😑

ChatGPT or equivalent is notoriously biased to certain cultures and demographics, therefore if truly embracing independent academic study you have to be careful it isn't baking in assumptions into your work, missing important elements of the evidence base and/or preventing you from doing important critical thinking (let alone all the other valid concerns raised already). You have to remember it's outputs are probabilistic and highly reliant on the data used to train it on.

But I do use it to cut down time on things where starting from scratch is a pain. Or to tidy up something fairly mainstream. This is no different to Googling for templates and ideas etc. Lecture ideas, workshop tools, framing particular well known topics, writing policies etc. All are sense checked and tweaked before hitting the real world though as the quality from AI isn't that great and it definitely hallucinates.

I certainly don't put sensitive or research data into it.

I agree with this.

I don’t want to give away identifying information but in order to continue my research career, I’ve had to pivot to teaching on a vocational degree programme which I didn’t myself train in, and I’m under quite extraordinary time pressure due to a very heavy teaching load. My research is excellent but according to you I shouldn’t bother? Thanks for making a fellow academic feel absolutely rubbish about themselves 👍🏻

I've seen a few people submitting answers that used ChatGPT or another AI tool that made their answer look nonsensical....and bizarrely, they didn't seem to realise they had submitted something nonsensical. So they were rejected. (It didn't look like it was done by a human being like a friend or relative helping them, or simply the candidate being longwinded, because the answers were detailed, in almost encyclopaedic fashion, but clearly not answering the question.) They would have had a better chance of success if they had attempted to answer it themselves using simple sentences.