Eric MillerAug 22, 2023
Timnit Gebru (she/her).
And another message. There was absolutely no need for this and we got the companies wanting to create an AGI god, putting these students in this awful situation. These "AI detection" tools need to be banned.
Aug 22, 2023 at 1:26amWeb
Show threadSydneyAug 22, 2023

@timnitGebru This just sounds like when "plagiarism detection" tools hit my middle/high school, except in this case, there's no actual basis for its determinations.

Fantastic. The AI will just get more "human" and humans will just get "caught" more and more.

Show threadRaccoon🇺🇸🏳️‍🌈Aug 22, 2023
@sydneybrokeit @timnitGebru
I was unfortunate enough to be a very strong writer as a child, the kind of person who later went into writing technical manuals. I got flagged by those on a regular basis, meaning I had to spend even more time on my essays than everyone else, because I was constantly told to rewrite the offending parts.
Show threadSydneyAug 22, 2023
@Raccoon @timnitGebru Yep. It's about compliance, not learning, and it shows in how they implemented those goddess-awful systems.
Show threadjnyroseAug 22, 2023
@Raccoon @sydneybrokeit @timnitGebru yep. Also punished for reading too quickly, in my case.
Show threadJames M.Aug 22, 2023

@timnitGebru to make it even worse, the false-positive rate of these "AI detectors" is much higher for non-native English speakers:

https://themarkup.org/machine-learning/2023/08/14/ai-detection-tools-falsely-accuse-international-students-of-cheating

AI Detection Tools Falsely Accuse International Students of Cheating – The Markup

Stanford study found AI detectors are biased against non-native English speakers

Show threadRaccoon🇺🇸🏳️‍🌈Aug 22, 2023

@jamesmarshall @timnitGebru
That makes sense. The kind of writing that AI does, especially if you don't lead it into any particular style, tends to be that sort of dry read that non-native speakers end up writing, with similar mistakes, possibly because it's been trained on other non-native speakers making those same mistakes: I'm certainly impressed at how well these supposedly English AI can handle switching into Spanish or German for me. I also wonder if the fact that translation software is now AI driven means similar patterns will show up.

Once again, minorities get disproportionately affected by these Tech Bros.

Show threadRaccoon🇺🇸🏳️‍🌈Aug 22, 2023

@jamesmarshall @timnitGebru
Actually, you know what also makes sense? They are using cheap labor in the third world by hiring non-native speakers to generate example text to train these things off of, and those people are writing their prompts in a way common for non-native speakers. Another theory would be the point out that most text is probably written by people who aren't from the main Anglo-Sphere of the US, Canada, and Britain.

I'm just fascinated by the number of problems with this entire model.

Pretty sure though that part of the grift here is to make a text AI, claim that it can write like any human, then get you to buy their software specifically made to catch their AI. Whether or not the AI or the AI detection software actually works reliably is irrelevant, as long as they can get you to pay for it.

Show threadFrank Bennett 🇯🇵Aug 22, 2023
@timnitGebru +💯 These "services" are an abomination.
Show threadGOTV 2026Aug 22, 2023
@timnitGebru this written English isn’t very good. A 4 gpa? Smh
Show threadRaccoon🇺🇸🏳️‍🌈Aug 22, 2023
@banjopat @timnitGebru
They do a very good job of explaining the situation. For me, that would be enough to give a student a good grade, especially if it wasn't a class about writing. Aside from that, remember they are stressed and probably writing this in a hurry, at the end of their rope after being threatened with consequences: they may write better when it's an actual project, and they are paying attention and proofreading thoroughly.
Show thread✨buff dog himbo✨Aug 23, 2023
@banjopat @timnitGebru this is a fucking EMAIL, not a professional or master's-level writing assignment, jfc. This reads like an extremely clearly written, properly punctuated, non-professional email between contemporaries tbh. Edgelords trying to flex like they're using highly sophisticated, elevated, colorful writing in every text communication they ever send LOL.
Show threadEibrielAug 22, 2023
@timnitGebru And @mozilla is not helping https://foundation.mozilla.org/es/blog/how-to-tell-chat-gpt-generated-text/
¿Detector de Chat GPT? Cómo saber si un texto fue creado por una IA

Chat GPT se utiliza para redactar documentos, artículos de noticias e incluso respuestas de mensajes de texto. Aquí te mostramos cómo saber si el texto que estás leyendo fue escrito por una IA.

Mozilla Foundation
Show threadBruno RohéeAug 24, 2023
@eibriel @timnitGebru @mozilla considering the quote "Detector tools will always be imperfect, which makes them nearly useless for most applications" actually they are?
Show threadspheruliticAug 22, 2023
@timnitGebru this is where you need to get a lawyer. Anyone blindly trusting these “tools” are certainly being negligent.
Show threadDebbie Goldsmith 🏳️‍⚧️♾️🇺🇦Aug 22, 2023
@timnitGebru I don't see that anything is going to fix this but a lawsuit. Perhaps a class action lawsuit.
Show threadK~Aug 22, 2023

@timnitGebru

I know AI detection was a hot thing for a brief moment; but my impression was that the brief moment has already passed - and people generally now understand that the AI detection is inherently unreliable.

Here is an article about why AI detectors claim the US constitution was written by AI:
https://arstechnica.com/information-technology/2023/07/why-ai-detectors-think-the-us-constitution-was-written-by-ai/3/

And here's an article about OpenAI discontinuing their AI detector because it doesn't work:
https://arstechnica.com/information-technology/2023/07/openai-discontinues-its-ai-writing-detector-due-to-low-rate-of-accuracy/

So yeah; people should stop using it.

#AI

Why AI writing detectors don’t work

Can AI writing detectors be trusted? We dig into the theory behind them.

Ars Technica
Show threadJoel Wirāmu, PaulingAug 24, 2023

@karadoc

@timnitGebru

Pre LLM I got accused of plagiarism for participating in a Newsgroup discussion on a Neuropsychology paper discussing a topic I was writing a report on. I can't explain how demotivating and disincentivising being accused of plagiarism is when you've actually put effort into primary research.

Show threadMartijn VosAug 24, 2023
My advice: use that same AI detection tool on stuff written by that teacher.