You are viewing a single comment's thread:
RE: The Stupidity of Hivewatchers
As a University professor, I’m dealing with students using ChatGPT to cheat on their essays on a weekly basis.
I wrote a post a couple months ago coming to the same conclusion as you — this is not a tool we can fight. We have a tiny edge right now because those using it are novices and don’t know how to cover their tracks. But the cover-your-tracks tools will soon appear.
As professors, we need to change our practices, which is something I am actively pursuing.
On Hive, we probably need to move to an engagement and genuine-reputation based system. Come and prove you’re a real person with real thoughts about real topics, and get rewarded in the process.
And that doesn’t happen by posting original content as much as it happens by interacting with real people, having real conversations.
In fact, that’s the essence of the Turing test. AI can’t pass the Turing test until it can fool most people during an ongoing dialogue.
I like the way @theycallmedan has been framing upvotes on CTT lately. Every upvote is a vote to help decentralize governance. Focus on upvoting those who can be a net positive force toward maintaining account level censorship resistance.
The adjustment to educational practices is kinda what should be happening anyway, it's just kind of hard and labour intensive to do so: Show your working.
A timeless curse of mathematics classes, something which would reliably filter out the fakes even in our art classes and, if it was possible yet, in my own music classes. The journey is, after all, much more valuable than the outcome @tarazkp ;)
It may be the case that ultimately, students will have to have office time with professors to demonstrate their human capabilities when compared with their submissions. I can imagine this being rather torturous if you have like 1,000 students, but there will be ways to implement it, i'm sure.
To me, simply handing in an essay or dissertation in Uni was never enough to begin with
I remember one of my sons struggling with geometry proofs because he was so used to doing extensive math calculations in his head. He just couldn’t slow down enough to write down every step.
He once failed a multiple choice chemistry test where he got 96% of the answers correct but did not heed the instructions to “show your work.”
That requirement can be brutal for kids like him.
Even so, simply learning to follow directions is an important life skill.
It is hard to fake when you have to show every step.
View more
Hey Steve, you must check out this article:
https://www.zdnet.com/article/this-professor-asked-his-students-to-use-chatgpt-the-results-were-surprising/
I found it quite interesting. A professor from the University of Pennsylvania deliberately asked his students to use Chat GPT to complete their homework.
Thanks for sharing!
Yes, I considered doing something like that. And I still might for future semesters.
We are still in very uncharted waters.
You probably are aware of the tool the chatgpt relesed for checks but just in case
https://platform.openai.com/ai-text-classifier
There are better tools, theirs is one of the worst (for now, I'm sure they will improve it).
View more
This is one of the tools that were are using to detect AI texts in articles on Hive.
View more
Yes. Thanks for pointing that out.
OpenAI’s classifier tool is pretty much worthless, though.
It only flags its own output 25% of the time. And it only takes slight modifications of its own output to fool the classifier even when it otherwise would catch it.
And, it’s easy for students to just keep making small changes then running it through the classifier until it’s “safe” to submit.
View more
The issue with the majority of these tools is that it is a blackbox. You cannot tell what it measures to determine if a text is ai written.
I fully support your POV as presented in the bottom half of your comment.
But vast majority of people discussing the HW issue stays in the original narrative of Hive blockchain needs a content so it distributes tokens to content producers.
Now the AI-generated content is like GMO-food. Some hate it on principle, others do not mind and may focus on various different parameters instead.
Ultimately, AI-generated content can be cornered on Hive if that's what the stakeholders end up enforcing but at this point anyone stating it is obvious AI content needs to be policed is either narrow-minded or brainwashed. Even taskmaster reminds us the YouTube grew on cat videos right under his post that takes the no-AI-rule for granted (while trying to grow Hive by finding the 2023 version of cat videos for/via LeoThreads).
Now, any reference to academical environment is thin ice. Universities have terms of service. Content creating on Hive does not.
Does any cat video consumer care whether they are manual or AI? Unless the AI origin is shoved in their face specifically to make them feel deceived?
HW was originally funded by DAO to detect stolen goods. People who dislike that tend to have hard time in most environments for a good reason.
Let's stay on topic. HW issue is about enforcing a subjective ruleset on manufacturing goods (which happens to be text/audio/video content here). Fun fact: the word literally refers to producing goods using [human] hands.
It is nowhere near the stolen content/food issue.
It has nothing to do with grow-your-penis spam that won't destroy the chain as any human reader can easily DV it (despite taskmaster telling us otherwise in his best gatekeeper voice).
Objecting the war on GMO-content on Hive is totally legitimate.