Wow, turnitin aren’t half misleading with that headline.
Makes it look like they have some bulletproof process. But what it means is they won’t flag something as likely unless their algorithm ranks it as 98% likely (proxy to limit false positives I guess). Not that they’re 98% accurate which is how the headline makes it sound.
They don’t seem to show any figures for how accurate it is at identifying AI written text.
I’d argue it will very rarely catch one that meets that requirement, and even by their own disclaimer, it can’t on its own be relied upon to actually identify it.
Which is why a lot of universities aren’t using it until they have tested it themselves. It’s not like the university would have much evidence anyway. They can’t exactly kick out a student based on a tool that isn’t 100%. With normal plagiarism, you can at least often see where the student copied from.
So I thought I’d finally try this AI thing - registered for Dall-E only to find out that I missed the free credit boat by a week or so… any other stuff I should try that is free?
Grammarly now has chatgpt superpowers too. I can’t stress how tempting it’s been for me when I’ve been struggling with a specific sentence, to get it to rewrite the whole thing for me. Pretty dangerous here, and I’d honestly rather it be entirely separate from my grammar and tone checker, which exists purely to make sure my writing is free from tone and grammar issues.
This academic year is going to be very interesting indeed. Grammarly’s premium plans can also make sure your work won’t trip up that particular tool that claims it can spot AI written dissertations.
I’m not sold on his little device personally. It’s cool.
But the phone can serve the purpose here just as, if not more effectively.
I like the whole tech being invisible mantra, but it’s somewhat ableist.
In any case, the demonstrated potential of AI, does feel like the future of how we’ll primarily be interacting with the internet. It’s an existential threat for Google, IMO. There’s no avoiding it. This technology will kill the google search engine we know today. I truly believe that at this point.
Humane are one company to keep an eye on, that’s for sure.
I have jused used Google Bard a bit and it has massively improved. The information seems to be a lot more detailed and accurate. It also seems better layed out and explained that Bing AI. I also just got my first advert on Bing AI which looked terrible and was irrelavent to my question (was asking about tables of data, it showed an advert for actual tables). I’ll still happilly use Bing AI as it was already pretty good, but Google is definitely getting better.
So I tried to sign up for an OpenAI account again and remembered why I don’t use ChatGPT. It wants my phone number for some weird reason, and even when I give it, it goes into a loop of asking for my b’day, asking for my number, asking for an OTP and then repeat… dodgy as anything if you ask me. Why do they need my phone number and my email?
