Do ChatGPT Detectors Work? Let's Find Out!

In this video I explore and compare 3 different GPT-3 AI detectors and pick a winner.
🛠️ Explore hundreds of AI Tools: futuretools.io/
🐤 Follow me on Twitter: / mreflow
🐺 My personal blog: mattwolfe.com/
🌯 Buy me a burrito: ko-fi.com/matt...
🍭 My Backgrounds: www.futuretool...
Tools Mentioned:
www.futuretool...
www.futuretool...
www.futuretool...
Outro music generated by Mubert mubert.com/render
#AI #GPT3 #AIDetector

[@fenomenailesi]

The standard software solution for detecting AI generated images/deepfakes now is definitely Winston AI

[@brabus3295]

I’m a university student, and in a particular class, students are being accused by the associate dean of plagiarizing work. I took emails sent from the associate dean and pasted them into an ai detector, and it said ai generated. How should we respond to these accusations?

[@brabus3295]

Students are being sanctioned if they cannot prove that their work was original

[@tebohosefatsa8280]

@@brabus3295 But how do you prove that your work is original? AI-written text is mostly if not always the same as human-written text.

[@carultch]

@@tebohosefatsa8280 AI Bots need to keep records of the conversation on a blockchain, so you can detect their use, and so these detectors can actually work, and be admissible in court.

[@seniorp9444]

Very interesting and hoping you do more tests to see how the detectors deal with text that has been slightly edited. I don’t think any of these a reliable enough to meet any standard of proof that a text was or was not AI generated.

[@susiet6246]

Great job, Matt! I’m new to AI, and am looking forward to your followup video with additional tools!

[@KezbanSuat]

Great tutorial! This is exactly what I needed to get started.

[@DaphneMarina88]

pretty cool. I downloaded. It only has 2 *, and you showed why. Appreciate your work so much Matt. Thank you.

[@denisalvarez7354]

im writing a report and I copied like 2 lines in it from chat gpt. Now Im terrified of it getting detected!!Is it possible I can get into plagiarism trouble?

[@GS-po4uo]

And?

[@abelzoni2138]

Will an AI detectors hold up in court?

[@sozzysucks]

lol absolutely not. the only way they can prove that is something like embedded watermarks or proof of connection

[@martymcfly6411]

No. They are based on probability and are also trained on human models. If you ever get a failing grade from a false accusation then sue them

[@SaidThaher]

Good video... if its true that those detectors worked many ( free money earnings ideas using AI ) will be worthless ... But try this : copy the text which AI generated to a word document... make some changes or rewrite some words .. and copy it from there to the detectors And see .. Or maybe try another subject ( not wolf selling hitdogs :) or what ever .. try something human .. and see .. Waiting for update .. gl

[@mreflow]

Thanks for the tips. I'm already planning on making a followup video because I've come across more tools and want to test more variations. I'll play around with your suggestions as well in part 2. :)

[@kiwiscribe]

@@mreflow I'll look out for that. Someone suggested using the semi colon ; and British spelling (good for me as we use it here in NZ) to get around AI checkers.

[@BaverMustafa]

This video is a game-changer! Thanks for the insightful content.

[@vandaken3612]

The ones I tried are not accurate. I wrote a story from scratch using grammarly to edit. When I copy and pasted into the AI detectors, one said it was 64% human generated, the other said 54%

[@m.j.mcintear793]

Scary to rely on machines that say you did it when you didnt lol. hmmmmm one day

[@thisiselonai]

Stocks in Vaseline are going to be bullish. 😇

[@martymcfly6411]

They don’t work period. Ive taken writing i did back in the 90s and it says 70% AI written. I feel bad for students and authors, artist etc, who are going to be falsely accused of things they didnt do.

[@LeonardCunningham-r4n]

Speak too fast