To think it's very easy to spot AI writing?

[Apocketfilledwithposies]

Does anyone else notice this?

Its always the same sort of style of writing. I can spot it a mile off!

Facebook pages seem flooded with it lately.

I have used it at work as a draft text. I usually have a laugh at the stupid phrasing but it acts as a springboard for me to get started on writing something.

It can definitely help when you don’t know where to start. I’m terrible at writing my annual performance appraisal/self review thing. It just feels horribly cringy to write and I get stuck even starting. We have to get feedback from other people on our performance, so I give copilot the feedback I’ve had and ask it to write a self review based on that. What I submit in the end bears little resemblance to what the AI first draft was, but I find it much easier to work from that than a blank page.

I work for a consulting company and we are heavily moving across to using Chat for everything. Its wildly alarming and I think rather short sighted.

I don't see the issue. When writing letters of complaint or in professional correspondence, AI helps get my point across nicely. But I'm not the best writer. I'm not especially concise and can veer between too confrontational or too accommodating. I do give it cues though and check it. Interestingly, the 7,000 or so LinkedIn posts I see bemoaning this very subject all sound like they've been written by AI.

I'm looking for a new car at the moment and a lot of the adverts on Autotrader are obviously AI and painful to read. Three paragraphs of sales puff waffle about the make and model in general but barely anything about that actual car. But that's shit AI generated stuff, the good stuff is I expect virtually indistinguishable from 'real' writing, unless you read an awful lot of it.

I agree. I don’t buy an item on EBay if it has the AI generated description/waffle

Sometimes you can tell, but it's not always so easy, and to echo a pp, I've been using dashes for decades and will continue to do so. That's simply my natural writing style! Sorry if it reads like AI to some!

Incidentally, I use AI as a writing tool five days a week. It can be helpful for getting started, as others have noted. I'm well aware that it may eventually put me out of a job, but it's here, now, and we may as well benefit from it however we can, for as long as we can.

I can't tell at all.

But I'm not the best writer. I'm not especially concise

AI isn't always concise either. It's often repetitive.

I hate AI. Let it do stuff like spot diseases or fraud, don’t create fake videos, pictures or robotic writing

I posted this then completely forgot I'd made the post. Apologies. 🤦

It's interesting those saying often autistic people are accused of their work being AI as I had a uni lecturer say they were impressed with my assignment and they had checked my work wasn't actually AI!

I also appreciate the point that maybe I'm only spotting the badly generated stuff and the good stuff is getting past me. I'd not thought of that but it is possible I guess.

Most of the ones I'm noticing on Facebook are about a person/event/place and the post tells the story of it. They're all a very similar sort of style of writing, often accompanied by an AI generated image. Or a very long widened emotional story with the tedoo app sprinkled in to the story about 3/4 of the way down. 🙄

I care. I hate terrible writing and trust me, you can spot AI. In some contexts I agree it doesn’t matter - if I’ve just asked Google a question and need a factual answer. But anyone who values style and character in language will wince. And it does seem to create a waffley, bland style.

Reading AI posts always seems like a waste of time to me and over the past few months I’ve moved away from forums and social media because of it. Bad AI writing is easy to spot but sometimes I’m not sure if it’s AI or not, so I may as well not read any of it. I’ve mostly gone back to books and newspapers instead and don’t regret it.

I've been accused on here before as being AI .
Tbh , I didn't know whether to be outraged or flattered .

Do you really not? Because as an educator, who wants students to train and use their own brains, I do.

In some contexts I agree it doesn’t matter - if I’ve just asked Google a question and need a factual answer.

AI isn't always factual.

Very good point. I agree, but I was only referring to the language style. The content is another can of worms!

It will get better.

We issue corporate reports on the quality of things. They come across fairly dry because there are certain judgements/measures that you have to go into, in a short word count, and how many different ways are there to say things like "people are happy with the service" or "the leaders manage things well". So you can sprinkle in the odd direct quote or colourful example line but most read fairly similarly. The organisation wants to trial having AI do the first draft of these reports, to then be reviewed and edited by a human. I think it's a ridiculous idea.

Why would anyone care to read these reports if - basically - we can't even be bothered to write them? If you're told a human who visited this place, spent several days there, spoke to lots of people and reviewed lots of things, wrote it - then you may disagree with the findings but you know it's based on something. If in essence ChatGPT wrote the report using summaries fed into it, what is the bloody point.

Yeah I think I can always tell. It's not just the uncanny valley clinical tone but the way it's simultaneously good and bad (sentences constructed perfectly but not the ideas they contain, to be reductive). Humans usually give away their individuality with turns of phrase.

That said if a really clinical textbook, about a subject I didn't know much about, was written by AI, I probably wouldn't be able to tell.

Oh yeah, spotting AI writing is getting pretty easy these days. It’s like the robot who tries way too hard to sound smart but ends up sounding like your dad reading a manual. You get those weirdly formal phrases, repetitive stuff, and zero real personality - as if a toaster decided to write an essay. No guts. AI doesn’t have guts yet, so if it sounds too polite and bland, trust your instincts!

Oh god I hate that eBay introduced AI descriptions! I’m with you on that one

I think that’s a bit of confirmation bias. Like a PP said, it’s like when people say you can always tell who’s had plastic surgery because they’re just remembering the obvious ones. With good work, you don’t notice what you don’t notice, iyswim.

Same with AI. When something sounds off, people jump to “that must be AI,” but there’s also AI-written stuff that reads normally (or at least, normally enough to not stand out), so they never question it. You only spot the bad examples because the good ones don’t draw attention to themselves.

True dat. I'm on a OU course and tutors trying to spot AI is becoming quite contentious and full of false-positives.