I did a straight up double take and rewind 30 second when I heard your be reviewing court jester. Hell yeah! Honestly, you almost have to review all the fights in that movie to really do it proper justice
Not to be unduly complimentary, but this video actually really well demonstrated your skill at doing all of this. It simultaneously demonstrated how well you can read the AI script, and the deficiencies of the AI script in turn show the skill in crafting the normal script, then we have precise editing to make it all work. Though perhaps intended all in good fun, this video could well serve for a demo reel, if such thing is needed. Also: it was very entertaining.
I find it interesting that the ai generated script has far more context on who the character is, what their role is in the source media, and less detail on the armor than your normal review videos – I suspect because the data source pool it’s been trained on has more in-depth reviews/discussions from media fans about media or characters they’re invested in/diving deep on in a wholistic fashion than there are tier list style videos in your style/with the sourcing method you use (reviewing outfits prompted by audience members, from media you’re not necessarily familiar with before reviewing). It makes for a much more generic review, and the lack of accuracy scoring in the ChatGPT model (its generated text is scored for syntax and grammar/natural-ness of the wording, NOT on truthiness of the content) makes the completely wrong descriptions of what it’s “reviewing” a recognizable pattern.
All of which is to say, comparing and contrasting a machine learning model’s generated script to your usual content is a great way to highlight the unique details you bring to the videos you make. I’m gonna ramble a bit more in a reply to my own comment, but overall: thanks for sharing the experiment, and hope you have fun continuing to make amazing videos!
ChatGPT is really stupid. I ask it two questions. 1. How bright is the Moon's surface in lumens? Answer. 111,438 lumens. 2. Why didn't that light show up in the Moon landing footage? Answer. The light level was too low to be picked up by the camera.—- So ChatGPT thinks that 111,438 lumens is low light level. It also thinks that space suits can be pressurized. Like i said really stupid.
i wrote a fantasy novel. never finished it, it's on my list of things to get back to. close to the end, but several chapters yet to go.
just looked. currently at 440,000 words.
…
… i think i mighta done something wrong.
…on a serious note, since there are technical aspect of things i know i got wrong, would you have a list of web resources for fantasy/medieval practices/technology?
i mean, that you can share easily? i ask because you seem to be an expert- at least a youtube expert- in it.
Ho no ! The IA can't replace us in our for fun activities yet ! What I nightmare ! I thought we could all forget the very concept of creative lesure activities and get back to our boring jobs, like in that book, 1984. What a shame !
I'm a writer and I was worried for my job security for exactly 5 seconds. That's how long it took chatgpt to come up with a 750 word Blog on a technical topic. It was 20% correct information, 30% word repetition and 50% BS
i wish chatgbt was available when i was in high school. it can do the brunt work of writing my essays and all i have to do is use a thesaurus and fact check everything lol
ChatGPT is pretty much the AlphaGo of BS. It will beat any human at its game, but in the end, it's still BS. (They're getting better at a quite alarming rate, though. GPT2 read like a mediocre high school homework, GPT3/ChatGPT generally has decent style and avoids most obvious errors, but tends to make up the details, and GPT4 already ranks in the top 10% at the bar exam and would win the Biology Olympiad.)
While this was a fun romp, was I the only one counting examples? I only counted four when it specifically stated 5 examples: Terminator, Matrix, Lara Croft, and Force Awakens.
No worry about AI uprising until Skyrim becomes sentient. Yes Skyrim because it's already installed everywhere… This video was stylish, functional and practical entertaining. Would wear it too if i had a display necklace.
Even when the AIs can make better content than this, it's best to give them a pass for moral reasons. AI always pulls from actual artist's content (without their consent) to create their work whether it's visual or written. This one in particular made a conglomerate of Jill's videos.
"I'm suddenly less concerned for my job security." As a Whump artist, this is how I felt once I realized that ChatGPT refuses to produce anything more dramatic or non-consensual than (literally) a Tickle Fight. I managed to get it to write a tickle fight between Winnowill (ElfQuest) and Loki (MCU) in which Winnowill got Loki to follow her commands because he couldn't take any more tickling, but that was literally as "dark" as I could get it; anything worse and it refused. (I'm not even talking anything gory or sexualized, just like having a villain force a character to do something they don't want to do — it's too "nice" to follow such prompts, and lectured me a bit over the idea as well.)
(Plus, it didn't like the idea of writing about Vampires, so how good could it possibly be?)
I did manage to get it to write some interesting crossover scenarios — once I primed it to actually figure out which characters I meant — but the thing I noticed is that it's like a rough draft, or a rudimentary children's story, and fails to hit that emotional heart that pulls you into the action. Telling, not showing, not delving into the experience. Plus, it can pull out some surface-level compare-and-contrast elements, but nothing really deep or particularly surprising.
And its understanding of Loki as a character is basically a "But Not Too Bad" Card-Carrying Villain caricature who just needs a little nudge to realize he'd fare better if he gave up his evil ways, which… is not in the slightest how I, or most of the fic writers I'm familiar with, perceive him. (I did have fun getting ChatGPT to pair him up with the likes of the caveman Cro, or XR from Buzz Lightyear of Star Command (the original cartoon).) So it's all very surface-level, and not going to replace fiction writers (especially fic writers) any time soon.
The main fallout for me personally, so far, is that I've twice received spam comments claiming that my podfics were AI-generated. Since one of them is my nephew playing GLaDOS, I took this as a weird sort of compliment!
I wear a pair of flip flops when I go do my shopping. They’re practical and functional with a looks that’s both stylish and practical. Style: 10/10 Functionality: 8/10 Tier: Survival Ready
The scary thing about AI is that you actually have to have some idea about what you're talking about to know that it's wrong. I know more about cheese than I do women's footwear and I would probably believe everything the AI said if you weren't correcting it (boots being kind of hard to see when casually watching a video).
You should consider having a one time $1 payment on Patreon just for people who want to read the first draft, just so you get a little something, don't TELL people how to get it for free.
Im so dissapointed the AI created the concept of "tank mode" on the armor tier list and then never tried to use it, that'd be like choosing combat wedges over combat boots. For a stylish, practical, and functional look of course.
I do hate to disagree, but I think that there is a real problem here. The important thing is not that ChatGPT was unable to produce the finished article. The important thing is the difference in the worktime required to produce the finished script starting from scratch, compared to time it would take starting from the ChatGPT script.
Suppose I ran a magazine or a TV show and I wanted an Jill Bearup slot. (And why wouldn't I?) I could hire Jill Bearup and pay her for the time it would take her to research and write this excellent content. Or, I could get ChatGPT to knock out a load of scripts like this for nothing and then pay somebody much less money to spend much less time knocking them into shape. And some of them would be quite good.
And, all of this would only be possible because of the work that Jill Bearup done and made available. And she could not demand any pay or even acknowlegement. This is not cool.
Turns out we hit 500k last night! ❤❤❤ Thank you all so much!
But are they practical?
I did a straight up double take and rewind 30 second when I heard your be reviewing court jester. Hell yeah!
Honestly, you almost have to review all the fights in that movie to really do it proper justice
Not to be unduly complimentary, but this video actually really well demonstrated your skill at doing all of this. It simultaneously demonstrated how well you can read the AI script, and the deficiencies of the AI script in turn show the skill in crafting the normal script, then we have precise editing to make it all work. Though perhaps intended all in good fun, this video could well serve for a demo reel, if such thing is needed. Also: it was very entertaining.
I find it interesting that the ai generated script has far more context on who the character is, what their role is in the source media, and less detail on the armor than your normal review videos – I suspect because the data source pool it’s been trained on has more in-depth reviews/discussions from media fans about media or characters they’re invested in/diving deep on in a wholistic fashion than there are tier list style videos in your style/with the sourcing method you use (reviewing outfits prompted by audience members, from media you’re not necessarily familiar with before reviewing). It makes for a much more generic review, and the lack of accuracy scoring in the ChatGPT model (its generated text is scored for syntax and grammar/natural-ness of the wording, NOT on truthiness of the content) makes the completely wrong descriptions of what it’s “reviewing” a recognizable pattern.
All of which is to say, comparing and contrasting a machine learning model’s generated script to your usual content is a great way to highlight the unique details you bring to the videos you make. I’m gonna ramble a bit more in a reply to my own comment, but overall: thanks for sharing the experiment, and hope you have fun continuing to make amazing videos!
Just tell chatgpt that she should use more synonyms for stilish . Then you get a little more chanced text. 👍🏻👍🏻👍🏻
Thanks for this!!! You made me laugh out loud today.
ChatGPT is really stupid. I ask it two questions. 1. How bright is the Moon's surface in lumens? Answer. 111,438 lumens. 2. Why didn't that light show up in the Moon landing footage? Answer. The light level was too low to be picked up by the camera.—- So ChatGPT thinks that 111,438 lumens is low light level. It also thinks that space suits can be pressurized. Like i said really stupid.
60,000 words.
sixty.
thousand.
words.
i wrote a fantasy novel. never finished it, it's on my list of things to get back to.
close to the end, but several chapters yet to go.
just looked.
currently at 440,000 words.
…
…
i think i mighta done something wrong.
…on a serious note, since there are technical aspect of things i know i got wrong, would you have a list of web resources for fantasy/medieval practices/technology?
i mean, that you can share easily? i ask because you seem to be an expert- at least a youtube expert- in it.
Ho no !
The IA can't replace us in our for fun activities yet ! What I nightmare !
I thought we could all forget the very concept of creative lesure activities and get back to our boring jobs, like in that book, 1984.
What a shame !
(Sorry, I had to ^^)
I'm a writer and I was worried for my job security for exactly 5 seconds. That's how long it took chatgpt to come up with a 750 word Blog on a technical topic. It was 20% correct information, 30% word repetition and 50% BS
"Functional" and "practical" are synonyms. It's like saying they're big and large.
I'm detecting a pattern here
I'm hype for the court jester video
i wish chatgbt was available when i was in high school. it can do the brunt work of writing my essays and all i have to do is use a thesaurus and fact check everything lol
ChatGPT is pretty much the AlphaGo of BS. It will beat any human at its game, but in the end, it's still BS.
(They're getting better at a quite alarming rate, though. GPT2 read like a mediocre high school homework, GPT3/ChatGPT generally has decent style and avoids most obvious errors, but tends to make up the details, and GPT4 already ranks in the top 10% at the bar exam and would win the Biology Olympiad.)
While this was a fun romp, was I the only one counting examples? I only counted four when it specifically stated 5 examples: Terminator, Matrix, Lara Croft, and Force Awakens.
You should repeat this in 6 months with whichever AI model is the fanciest at the time… Unless we all dead or enslaved by it…
No worry about AI uprising until Skyrim becomes sentient. Yes Skyrim because it's already installed everywhere…
This video was stylish, functional and practical entertaining. Would wear it too if i had a display necklace.
i really like the boots that were stylish and practical
Take a shot everything she says practical
A.I. – “artificial idiot”
502K subs! WOOOOOOOOOOOOO!
Even when the AIs can make better content than this, it's best to give them a pass for moral reasons. AI always pulls from actual artist's content (without their consent) to create their work whether it's visual or written. This one in particular made a conglomerate of Jill's videos.
"I'm suddenly less concerned for my job security."
As a Whump artist, this is how I felt once I realized that ChatGPT refuses to produce anything more dramatic or non-consensual than (literally) a Tickle Fight. I managed to get it to write a tickle fight between Winnowill (ElfQuest) and Loki (MCU) in which Winnowill got Loki to follow her commands because he couldn't take any more tickling, but that was literally as "dark" as I could get it; anything worse and it refused. (I'm not even talking anything gory or sexualized, just like having a villain force a character to do something they don't want to do — it's too "nice" to follow such prompts, and lectured me a bit over the idea as well.)
(Plus, it didn't like the idea of writing about Vampires, so how good could it possibly be?)
I did manage to get it to write some interesting crossover scenarios — once I primed it to actually figure out which characters I meant — but the thing I noticed is that it's like a rough draft, or a rudimentary children's story, and fails to hit that emotional heart that pulls you into the action. Telling, not showing, not delving into the experience. Plus, it can pull out some surface-level compare-and-contrast elements, but nothing really deep or particularly surprising.
And its understanding of Loki as a character is basically a "But Not Too Bad" Card-Carrying Villain caricature who just needs a little nudge to realize he'd fare better if he gave up his evil ways, which… is not in the slightest how I, or most of the fic writers I'm familiar with, perceive him. (I did have fun getting ChatGPT to pair him up with the likes of the caveman Cro, or XR from Buzz Lightyear of Star Command (the original cartoon).) So it's all very surface-level, and not going to replace fiction writers (especially fic writers) any time soon.
The main fallout for me personally, so far, is that I've twice received spam comments claiming that my podfics were AI-generated. Since one of them is my nephew playing GLaDOS, I took this as a weird sort of compliment!
Ai cn never match Jill's humor😁👍. You're safe O Mighty One😊
I wear a pair of flip flops when I go do my shopping. They’re practical and functional with a looks that’s both stylish and practical.
Style: 10/10
Functionality: 8/10
Tier: Survival Ready
"Adjective word salad"
"Not 'falling outside the fetish club ' high"
You're a very quotably hilarious human.
I think there's still space for an AI Jill tier review of cheese. And how that tier list handles the idea of stabbing I'm just dying to know!
The scary thing about AI is that you actually have to have some idea about what you're talking about to know that it's wrong. I know more about cheese than I do women's footwear and I would probably believe everything the AI said if you weren't correcting it (boots being kind of hard to see when casually watching a video).
I love your videos – they are so stylish and practical!
This video is both practical and functional for style and practicality.
Men are totally prepared for the experience (If it's a female chatbot). They never listen to females anyway :p
Please include Tank Mode going forward. Please.
Did anyone mentioned for You to review templar armor From Hellgate: London ?
You should consider having a one time $1 payment on Patreon just for people who want to read the first draft, just so you get a little something, don't TELL people how to get it for free.
So much word salad. So, so much.
Actually u just have t9 change a few words
Im so dissapointed the AI created the concept of "tank mode" on the armor tier list and then never tried to use it, that'd be like choosing combat wedges over combat boots. For a stylish, practical, and functional look of course.
I'll be honest, I expected AI Jill to glitch out
That Tech-Rex shirt looks so stylish, practical and comfortable!! 😄 And also stylish and comfortable 😉
I do hate to disagree, but I think that there is a real problem here. The important thing is not that ChatGPT was unable to produce the finished article. The important thing is the difference in the worktime required to produce the finished script starting from scratch, compared to time it would take starting from the ChatGPT script.
Suppose I ran a magazine or a TV show and I wanted an Jill Bearup slot. (And why wouldn't I?) I could hire Jill Bearup and pay her for the time it would take her to research and write this excellent content. Or, I could get ChatGPT to knock out a load of scripts like this for nothing and then pay somebody much less money to spend much less time knocking them into shape. And some of them would be quite good.
And, all of this would only be possible because of the work that Jill Bearup done and made available. And she could not demand any pay or even acknowlegement. This is not cool.
I have actually found AI to be really good at editing existing information. It’s just terrible at coming up with stuff on its own.
That’s it. Practical doesn’t sound like a word to me anymore.
Congratulations on finishing your first draft.
Comments are closed.