Steve PeppleJan 31, 2023
Will Oremus

OpenAI's new tool to detect AI-written text correctly spotted 26% of AI-written text, while falsely identifying 9% of human-written text as AI-written.

One way of interpreting this: AI isn't yet good enough to reliably identify AI content.

Another: The technology to detect AI text is lagging well behind the technology to produce it. And/or it turns out to be a harder problem. https://openai.com/blog/new-ai-classifier-for-indicating-ai-written-text/

New AI classifier for indicating AI-written text

We’re launching a classifier trained to distinguish between AI-written and human-written text.

Jan 31, 2023 at 9:37pmWeb
Show threadShomilaJan 31, 2023
@willoremus Agree - It’s always been a hard problem for humans to detect if text/chat is human or AI e.g. Turing test. Maybe it’s impossible if OpenAI succeed in their first goal - to make the answers really convincing to a human? (Assuming that was their goal!)
Show threadSturmUndDrangerJan 31, 2023

@willoremus

Breaking: Microsoft announces that Windows 2049 will issue with a fully-integrated Voight-Kampff test.

https://nautil.us/the-science-behind-blade-runners-voight_kampff-test-236837/

The Science Behind “Blade Runner”’s Voight-Kampff Test

Is Rick Deckard a replicant, an advanced bioengineered being? The jury concerning the character in 1982’s Blade Runner is still out. Harrison Ford, who plays Deckard in the film, thinks he’s human. Ridley Scott, the film’s director, is adamant that he’s not.* Hampton Fancher, the screenwriter for the original film and the sequel, Blade Runner […]

Nautilus
Show threadJames WellsJan 31, 2023
@willoremus
So I am reading that to say it is wrong 35% of the time, so the same as any high school or college teacher.
Show threadTux 🪿Feb 1, 2023

@willoremus reverse Turing tesrts, here we go.

We all got userd to filling up endless capchas to prove wre klinda hiuman.

Just wait until yoy have to ace a humanitty yest severals a ddy

Show threadLeon Derczynski 🏔️✍🏻🌲☀️Feb 1, 2023
@willoremus slightly surprised they released it this early
Show threadΩ 🌍 Gus PoseyFeb 1, 2023
@willoremus Also, AI is a kingdom of lies.
Show threadMarcel SalathéFeb 1, 2023

@willoremus It is practically useless, for reasons outlined here: https://digitalepi.substack.com/p/how-useful-is-openais-new-ai-detection

See also https://mastodon.social/@marcelsalathe/109787200129188289

How useful is OpenAI's new AI detection tool?

As we've learned during the early phase of the COVID pandemic, detecting rare things requires accurate tests.

Digital Epidemiology
Show threadJesseFeb 1, 2023
@willoremus I feel like these are all the same interpretation?
Show threadDomFeb 1, 2023
@willoremus interesting. We keep debating where this will take us at work. Will we be able to make use of these tools. I keep saying that it will kill innovation and design thinking. Development roles will stagnate if we turn to tooling to write the code for us. Absence of innovative thinking and imagination could be the death of us. Be interesting to see what happens. Sentient AI on the other hand could be very useful and or very dangerous.