To get more direct to the point you could use those unrendered dummy links to ban whatever IPs click them.
With the vast amounts of training data and how curated they’re becoming (Llama and Claude are going that direction) it’s infeasible to actually poison a large model to this degree.
Oh yeah, Jezza seems great, from what I know he actively left the Mormon faith and was ostracized by most his family. He uses this experience to be a more empathetic and better person.
All the more sad when Shad tries to claim he’s just as good an artist and Jezza doesn’t have the heart to call him out on it.
Oh it gets worse with Shadiversity. Huge AI art guy, his brother’s an actual artist too so it’s hard seeing Shad brag to him. Very “anti-woke” and paints his conservative Mormon beliefs on everything.
The worst unforgivable part is the end of his book has impregnated rape victims step up to defend the rapist protagonist because he “gave them” a child, while the ones that didn’t get pregnant were jealous.
He loves to bring up that the book is supposed to explore this immoral character. But this isn’t the protagonist’s viewpoint this is just how Shad thinks the world works. This is how Shad believes rape victims think.
Very sad to see, I followed him for swords and castles but Jesus Christ.
There was a brief window after ChatGPT when I had a twinkle in my eye and thought “this is an impressive start, how will they improve it? How will they make it more efficient?”.
That went away when all the tech companies in unison slurred out “make it bigga!” then pushed to production.
It’s a little funny how everyone sobered up from perpetually investing in unprofitable free social media then they dove right back in to perpetually investing in LLMs with no real plan for sustainable profit.
In terms of LLM hallucination, it feels like the name very aptly describes the behavior and severity. It doesn’t downplay what’s happening because it’s generally accepted that having a source of information hallucinate is bad.
I feel like the alternatives would downplay the problem. A “glitch” is generic and common, “lying” is just inaccurate since that implies intent to deceive, and just being “wrong” doesn’t get across how elaborately wrong an LLM can be.
Hallucination fits pretty well and is also pretty evocative. I doubt that AI promoters want to effectively call their product schizophrenic, which is what most people think when hearing hallucination.
Ultmately all the sciences are full of analogous names to make conversations easier, it’s not always marketing. No different than when physicists say particles have “spin” or “color” or that spacetime is a “fabric” or [insert entirety of String theory]…
On Discord though there’s a lot of unchecked predation. Theoretically if this were implemented it would let them see the most suspicious users that interact with an unusual amount of children and review if the messages are inappropriate.
But all that’s unlikely because if they actually cared they’d implement other simpler solutions first. So this idea is just hypothetical but not ideal.
I’m a bit annoyed at all the people being pedantic about the term hallucinate.
Programmers use preexisting concepts as allegory for computer concepts all the time.
Your file isn’t really a file, your desktop isn’t a desk, your recycling bin isn’t a recycling bin.
[Insert the entirety of Object Oriented Programming here]
Neural networks aren’t really neurons, genetic algorithms isn’t really genetics, and the LLM isn’t really hallucinating.
But it easily conveys what the bug is. It only personifies the LLM because the English language almost always personifies the subject. The moment you apply a verb on an object you imply it performed an action, unless you limit yourself to esoteric words/acronyms or you use several words to overexplain everytime.
The gender thing is creepy, but if they could predict age groups then in a perfect world they could analyze adult users talking to children and shut that down.
In a perfect world though, I doubt they’d put effort into making their app safer, heavens no.
I hate that it’s links are “incompatible” with Firefox, even though if you trick it into thinking it’s Chrome, it works just fine.
Can’t wait for FOSS brain implants, it would still be hellish but a fun kind of hellish.
I want someone like Linus Torvalds to verbally abuse someone for not understanding basic computational neuroscience.
Yeah I love Foundry, but I’m convinced the DM needs technical knowledge to use it. I ran a server for non tech savvy DM and it was like working customer service.
With plenty of investment you can get the tabletop to be almost exactly what you want it to be, and for a popular system like 5e you can make it as automated as a Baldurs Gate game. You just need to download a lot of modules to get there and customize a lot of settings. Without that it just becomes a less intuitive Roll20.
And I must stress from experience, never offer to host/troubleshoot a server for someone else, especially if the DM likes to complain or can’t handle minor technical setbacks.
I’m curious, is there actually so many 42’s in the system? (more than 69 sounds unlikely)
What if the LLM is getting tripped up because 42 is always referred to as the answer to “the Ultimate Question of Life, the Universe, and Everything”.
So you ask it a question like give a number between 1-100, it answers 42 because that’s the answer to “Everything”, according to it’s training data.
Something similar happened to Gemini. Google discouraged Gemini from giving unsafe advice because it’s unethical. Then Gemini refused to answer questions about C++ because it’s considered “unsafe” (referring to memory management). But Gemini thinks C++ is “unsafe” (the normal meaning), therefore it’s unethical. It’s like those jailbreak tricks but from its own training set.
deleted by creator
I think one of the main issues is the matter of fact usage of the term Minor Attracted Person. It’s a controversial term that phrases pedophiles like an identity, like saying Person Of Color.
I understand wanting a not as judgemental term for those who did no wrong and are seeking help. But it should be phrased as anything else of that nature, a disorder.
If I was making a term that fit that description I’d probably say Minor Attraction Disorder heavily implying that the person is not ok as is and needs professional help.
In a more general sense, it feels like the similar apologetic arguments that the dark side of reddit would make. And that’s probably because Google’s officially using Reddit as training data.
This is typically why education and experience are still needed if you’re self taught.
I know from learning programming that people online don’t explain “common sense” problems. So many times you’ll look up a problem and see people talk about huge refactors or complex niche fixes when in reality you misplaced a single line of code.
People don’t understand that I choose every word carefully in a sentence to convey the most meaning and answer follow-up questions up front. Then they think I said something that literally contradicts what I said or that I already accounted for in my first sentence.
Yes I’m a programmer with Autism and ADHD, why do you ask?
Trans, pan, 196, and Blahaj is friend.
I’m early Gen Z with a kinda poor family. So I had CRT’s and old VHS but also grew up on the internet.
I feel an extreme gap between me and people a few years younger. I graduated in 2018 so I was some of the last people to have a traditional highschool experience. Before Covid, Zoom, and Chatgpt.
I also mostly grew up with computers instead of phones so Im only just now getting into TikTok, I’ll likely never truly revolve around it like many others (both older and younger than me).