Or electronic calculators. It’ll take time for the TurnItIn type AI checkers to get good enough to check.
In the meantime, I’m just using AI to grade papers in courses where I can’t control student’s assignments and I expect at least some of them are using AI. It’s only an issue in a few shared courses so far, and actually the AI gives more feedback (and better and faster)
Methinks your AI does not account for creative language distortion, yet.
This is the exact plot of the South Park episode about AI!
AI text detectors are not reliable and will never be. There are a huge paint for student who don’t use AI because of the false positive. And they can be easily bypassed by cheaters.
Yeah, I’ve read this and believe it
Never is a long time
Um, what? I’m actually teaching students to use it in some contexts, so they can provide better writing. I just tell them to keep the first draft will all the mistakes, and then submit something that is easy to read after AI editing. Students who can show multiple drafts and stages of a developing work don’t need to worry about false positives.
I think good teachers are adapting to the AI, and teaching how to use it productively and responsibly.
Bad students forget to delete the identifiers that make it clear something was AI generated 
They don’t forget they just don’t bother
Not all professors are like you. Many universities stop using AI detectors. Is really bad when you submit your essay and the automatic detector is flagging it 100% AI. Some professors don’t understand the limitations and you may have to go to a process of proving your work. Is easy to shared the docs history but they may also say you use AI lol.
It is cheating IMO, unless the client has agreed to that and you’re being extremely careful to check the output and ensure an accurate translation and decent-quality work. I remember you having problems with clients before.
Computer/AI-assisted translation/writing/editing is one thing, but when a client is paying for human translators/writers/editors they do tend to want humans playing a major role. If they can get a similar result by just sticking the source text through a computer and not having to pay a human, you’re kind of shooting yourself in the foot in the long run by making yourself dispensable.
I think quite a lot of clients in this sector have already decided they don’t need humans for this step (or at all) because AI is “good enough”.
I check the output for error and make corrections where necessary (they often are). It’s more machine assisted.
In this case the agency has unreasonable deadline. They’ve ignored my maximum throughput per day repeatedly and insists I do over 4 times that.
What’s your maximum throughput per day? Is it reasonable?
5000 Chinese characters per day in extreme cases but my normal throughput is 2500 per day. They gave me a 30,000 character document giving me 6 days to do them.
That seems on the high side, I agree. But if you need the work and don’t have much else to do…
And as a freelancer you can always say “no”.
Thing is I’ve done it that same way for many years and no one had any complaint about the work, then suddenly the client goes nuts over dialects or something (they seem not ok with me writing in traditional Chinese then changing it into simplified for some reason).
Before this they made no preference as to what dialects they wanted. Maybe it’s because of some government mandate?
Ultimately I complained to proz about this and I was finally paid. I lean towards the rejection being manufactured. I think the agency had fired a few pm.
Yeah, I remember the thread.
IIRC, you were using Google Translate or similar to convert from traditional Chinese to simplified Chinese without accounting for regional differences in usage. My understanding was (and is) that this is bad practice and I suspect the agency had at least a partially valid point here. Which isn’t to say that shitty agencies/clients don’t exist.
Anyway, this is getting a little off-topic.
Yeah, it’s true
So I started texting in Korean using ChatGPT. I asked it to teach me particular grammar. Well, I’m impressed. I can’t wait until it supports voice for an actual conversation.
It uses the honorific form. I’m tempted to speak to it in the familiar form and demand that it speak to me in the honorific form. I’m curious if I’ll get some pushback.
I’ll start exploring more later. So far, very cool.
