balajis
@balajis.eth
I want a simple open source JS text widget that automatically detects and rejects AI input. - It constantly keeps up to date with the latest models - As the user types, it displays the probability that the input is AI - If the user pastes in text, the AI probability increases - It flat-out rejects input if AI probability is >50%, by graying out submit - It doesn't show anything unless the AI probability is high Basically, a good slice of people now paste in AI slop because they're lazy, can't write, or don't realize how detectable it is. I don't want to manually detect it with my eyes, I want AI to do that and reject it prior to submission. Put the widget up at noaijs.com or a similar domain. Allow developers to subscribe for a hosted version, or even perhaps buy a memecoin (ha!) to support.
11 replies
11 recasts
115 reactions
vrypan |--o--|
@vrypan.eth
CAPTLHA: Completely Automated Public Turing Test to tell LLMs and Humans Apart (Could be Klingon) https://www.youtube.com/watch?v=YceE15VQYHE
1 reply
0 recast
8 reactions
Kieran Daniels
@kdaniels.eth
Ya hiring the last 6 months has been a learning experience. Hired multiple people and directly told them we aren’t hiring you to just only use AI to write and they both just used AI to write everything and hid it poorly. The good news is GPT has improved so greatly that it’s better than a junior writer now. With memory and reasoning it’s insane. This will come in sweeping waves for design and coding as well.
1 reply
0 recast
4 reactions
samg.base.eth 🎩🍖
@samgslastlife
RIP every normie I know writing resumes if this gets built 🤣
0 reply
0 recast
1 reaction
yesyes
@yesyes
Is it even possible to detect and reject AI input reliably? I doubt that. Yes, AIs can have a certain way of saying things but that can easily be altered with a few clever inputs.
1 reply
0 recast
3 reactions
Vova 🎩
@vova
Maybe it’s time to rethink application processes that rely on text submissions. AI slop is not going anywhere and no automatic detection will catch all of it.
0 reply
0 recast
0 reaction
Mir
@mirsol
Great
0 reply
0 recast
0 reaction
JoshD
@joshdavis.eth
Good luck existing in the world over the next decade.
0 reply
0 recast
0 reaction
Andrei O.
@andrei0x309
There isn't a reliable way to detect AI as the goal of LLMs is mainly to reproduce language, as naturally as a human will do it. There's a ComputerFile clip that details how a model could in practice hide signatures inside the language outupt, but that has many drawbacks including decreasing the quality of the output. And you can't force model creators to do that anyway. This phrase "it constantly keeps up with the latest models" IMO just shows a poor understand of how LLMs, work.
0 reply
0 recast
0 reaction
patxol
@patxol.eth
If the code is open sourced, then it’s useless.
1 reply
0 recast
0 reaction
na
@na
- It constantly keeps up to date with the latest models. MCP Clients can subscribe to updates for specific resources: https://modelcontextprotocol.io/docs/concepts/resources#content-changes
0 reply
0 recast
0 reaction
KMac🍌 ⏩
@kmacb.eth
I wonder how cliches and platidues will score when typed. proximity scores probably high 🤔 Let's not feed the meme coin snipers when you can feed the network of contributors revnet.eth.sucks/memo
0 reply
0 recast
0 reaction
