So hes been producing high quality educational videos since ky childhood but now suddenly its too much pressure to do without AI?
And hard agree on the cope comment: how does someone who claims to love science insist AI isnt harmful? Even if you dont believe its gonna harm jobs, etc., have we all forgotten about global warming?
Lastly, I hate when people say its a research tool. No its not. Research is heavy mind activity and requires questioning what you are consuming - which is not what people using AI are doing. Theyre just copy pasting shitty incorrect graphs. Thats not research.
I partially disagree with you, but just on the research bit.
In the situation you describe, where people use AI for “research” and then just use and copy paste the output, I agree that's not research.
However, I've used it occasionally at work to get an overview of sources in fields I'm less familiar with, as a first pass to get some initial sources to dig into and some terms to then investigate further with google and wikipedia.
To be clear, I’m not an AI enthusiast by any means, I'd cheer if all AI data centers imploded in on themselves, but your comment is quite reductive of the way people use it. I could frame traditional modern research using a search engine the same way, saying people just copy paste whatever the first link to show says.
Thats not how traditional research works, at all, by principal or practice. If you are googling and just copy pasting the first link, thats also not research. But thats the thing, most people who are actually doing good research arent doing that. People have always been lazy, its just that now those lazy people can use AI and delude themselves into thinking theyre doing good work.
Honestly in my opinion, trusting a "source" that consistently produces hallucinations and gives you results it predicts you might prefer is not a good avenue for even the most basic research and that just keeps getting proven time and time again when all these respected people get sucked in and then poop out shitty incorrect work that they didnt actually check.
In my research I spend a lot of time reading papers and AI serves as a much better search engine than google because it can tell you what key words you should be searching for. I’m able to sift through the literature/determine what’s relevant much more quickly than I could if I were just clicking links. I’m able to pull on broader domains of knowledge than I could if I were confining my searches to the terms I already know. Lit review is a deeply important and extremely traditional part of any research. 50 years ago this meant going into a science library and reading titles of papers off a physical catalog. I’ll take AI over that any day.
yeah, this is the part of research I was alluding to where I do see a value of LLMs, as long as you spend the time reading those actual sources vs just copying or regurgitating what the AI is saying. As Smalki said, once you're just copying what the LLM is saying, you lose a vital part of what makes research research, in my opinion.
what I find difficult about this though is, the distrust I feel towards the LLM itself and its inherent biases based on biases in the training data and the people that supervised this process
I think what they’re talking about is just for learning what a source might be, and then going and looking up that paper. In that sense it’s acting like just a very good search engine, and yes it saves you from having to look through millions of less relevant papers, but I don’t think it really dilutes much academically.
Meanwhile 90% of other video essayists are running all their scripts through chatgpt and it's so fucking obvious if you have the truesight of ever using these tools yourself, it drives me mad every time I go on youtube. And yet nobody notices, while Hank gets crucified for using it to… collate research sources? Moronic.
“Unhealthy AI use” holy shit yall are not living in reality. The entire thing is a non story. And the author of this article clearly didn’t do enough digging because the “pushback” line is already explained.
Using “I appreciate the pushback” once isn’t evidence of relying on AI. AI use such phrases often because people do, hence it’s in their training data a lot. If someone’s work is riddled with it, then yes it’s probably AI. But the anti-AI crowd seems to be on a witch-hunt.
I think you can read an implication from the dopamine hit comment that he is not using the chatbots merely for research. As you noted, a lot of people find themselves in an unhealthy, sometimes even physically dangerous, feedback loop with these tools -- which are, as yu note, designed to keep people engaged. Combined with his family's alarm, I suspect he is actually caught in one of those feedback loops, using the bots for far more than just research. It is certainly plausible, given the information in your post.
Every time I see a post like this, I discover a new AI pattern that I use naturally in human speech. I’d so fluidly say, “I appreciate the pushback.” It’s disturbing to me that so many people are reading human thoughts and presuming AI usage when it’s not always the case. Feels a little like a witch hunt. I totally agree that AI presents dangers, but I do wish we could find a better way of handling it than our current posture of suspicion, cancelling, and condescension. Not a critique of you—just a reflection on what’s happening overall.
Yep. My entire point. AI “detects” certain phrases and patterns because it was trained on actual human writing. The AI detection system is just as goofed up as the generation system. When the public is told that a certain pattern indicates AI, they categorically dismiss what context should clarify. The em dash has been villainized. If/then statements have been villainized. Now we can’t use the word “quiet” or the phrase “Thank you for the pushback.” We need thoughtful people writing about this with nuance, not lonely Twitter bros working out wounded egos with childish comments and bravado.
This whole essay borderlines on victim blaming, there's a reason AI psychosis is a thing and not everyone has the tools to resist what some of this LLMs offer. If the guy says he gets a "dopamine hit" from continuous AI use I mean... he's not alone, we need more regulations for AI and that's the bottom line
As shocking as it is to see people publish something created by LLMs, and obviously discussing ethical and romantic conundrums, I don’t see why people get so mad about research with LLMs? You want me to waste time watching and reading a bunch of “fashion advisors” pretending to be emotionally invested in my sartorial choices? Or can I just use a chatbot for a quick starting point and then progress according to my individual interests?
"His line about “dopamine” hits from LLMs is silly and dramatic."
Hank Green was diagnosed with ADHD as a child. I was diagnosed with ADHD as an adult. I can tell you from myriad personal experience that using AI is an excellent way to get a dopamine hit, especially when you're in a conversation with it about a hyperfocus that bores the pants off the actual humans in your life. (It's one of the reasons I enjoy using AI for brainstorming, etc. It makes writing much more fun.) Is that healthy? Sometimes yes, sometimes no, and it looks like Green hit his limit of healthy interaction. But it's not silly or dramatic for him to say that.
Like Bill Nye didn’t have a team of dozens of people helping him, funded by the government 🤣 give the guy a break! It’s not easy to compete online at his level. The amount of content these creators have to push out, with their own funding for a team, is crazy.
And it's not the same, even for those that do. The picture of your plate on Instagram or your profile photo on some shit social site is not the same as an essay or a vlog.
My art is the product of forty plus years of study. You insult and you don’t even look. Not very bright are you? You’ve invalidate your own options straightaway. Is that the level of insight you bring to your writing?
You’re a bit of a daft twat or wanker, take your pick. My photography isn’t ai - i shoot with a Leica or an iPhone and I’ve never photographed a plate. Nor are my Antiselfies AI . If you’re getting this much wrong and being so judgemental and shouty how bad is any analysis in your writing? Highly shit I suspect. I certainly won't be looking at it. Get help.
I remember when The Fault in Our Stars came out. I was an eleven-year-old girl who curled up to read it under my peach polka dot duvet in the small box room that was meant to be a storage room. My wallpaper was duck egg green because my best friend's was duck egg green too. We posted our books on Instagram, back when it was vignette Pictochat filters, sleepover selfies, and showing off our new pumpkin mug with fairy lights in the background. We'd get 6 likes and we'd be happy. We'd talk about cigarettes being metaphors during recess, and then at home over Kik. We'd dip chicken nuggets in the vinegary sweetness of off-brand ketchup, read some more, and then go to sleep.
I remember my best friend with the duck egg wallpaper posting a photo of her crying at the cinema. Her and her other friends. They'd gone to watch the adaptation of pretty actors playing sick kids. They were holding blue slushies and wore chokers to try and look like cool teenage movie protagonists. We didn't know then that we were on the precipice of something dangerous--that those Pictochat filters would turn into air brushing, would turn into weight loss drugs that would kill some of the girls like us. We didn't know that, in ten years time, those 6 likes could be the end of a company, the focus of a life, the make-or-break of paying bills that month. We were just reading books next to fairy lights, drinking slushies and crying and going to sleep. We picked the wallpaper we wanted, not the one a YouTube manifestation guru told us to in order to "upgrade your chic". We trusted John Green to tell us things about mortality, empathy, and how people are people and not places. We felt and captured and thought about things. We spoke about them at recess, not just over Kik.
Soon, we'd be loosing the moment to a story photo--to a pre-meditated life plan, informed by robots and influencers and striking the perfect macros. I think we should go back to eating chicken nuggets for tea sometimes. I think we should put on some fairy lights, and curl up under a blanket, and face our mortality. We should use it a reminder to notice the duck egg wallpaper, and cry at movies, and to go to sleep at night, and not rely on men who forget what it is they once wrote, and sell away their empathy.
I am so disappointed in Hank Green. He really should know better. Hadn't his channel been active for years before the AI boom? He should not have been having any problems researching and scripting videos.😭
I work for a massive global corp (unfortunately) and it’s funny reading that recording meetings and plugging them into Claude is shocking because of how normalised it is at work. Where I work now, senior staff have Claude agents that work / converse with other agents on their behalf when they’re out of office. Recording meetings instead of taking physical notes is honestly the very tip of the iceberg! One of our KPIs is to use AI for 60% of tasks at minimum.
So hes been producing high quality educational videos since ky childhood but now suddenly its too much pressure to do without AI?
And hard agree on the cope comment: how does someone who claims to love science insist AI isnt harmful? Even if you dont believe its gonna harm jobs, etc., have we all forgotten about global warming?
Lastly, I hate when people say its a research tool. No its not. Research is heavy mind activity and requires questioning what you are consuming - which is not what people using AI are doing. Theyre just copy pasting shitty incorrect graphs. Thats not research.
I partially disagree with you, but just on the research bit.
In the situation you describe, where people use AI for “research” and then just use and copy paste the output, I agree that's not research.
However, I've used it occasionally at work to get an overview of sources in fields I'm less familiar with, as a first pass to get some initial sources to dig into and some terms to then investigate further with google and wikipedia.
To be clear, I’m not an AI enthusiast by any means, I'd cheer if all AI data centers imploded in on themselves, but your comment is quite reductive of the way people use it. I could frame traditional modern research using a search engine the same way, saying people just copy paste whatever the first link to show says.
Thats not how traditional research works, at all, by principal or practice. If you are googling and just copy pasting the first link, thats also not research. But thats the thing, most people who are actually doing good research arent doing that. People have always been lazy, its just that now those lazy people can use AI and delude themselves into thinking theyre doing good work.
Honestly in my opinion, trusting a "source" that consistently produces hallucinations and gives you results it predicts you might prefer is not a good avenue for even the most basic research and that just keeps getting proven time and time again when all these respected people get sucked in and then poop out shitty incorrect work that they didnt actually check.
In my research I spend a lot of time reading papers and AI serves as a much better search engine than google because it can tell you what key words you should be searching for. I’m able to sift through the literature/determine what’s relevant much more quickly than I could if I were just clicking links. I’m able to pull on broader domains of knowledge than I could if I were confining my searches to the terms I already know. Lit review is a deeply important and extremely traditional part of any research. 50 years ago this meant going into a science library and reading titles of papers off a physical catalog. I’ll take AI over that any day.
yeah, this is the part of research I was alluding to where I do see a value of LLMs, as long as you spend the time reading those actual sources vs just copying or regurgitating what the AI is saying. As Smalki said, once you're just copying what the LLM is saying, you lose a vital part of what makes research research, in my opinion.
what I find difficult about this though is, the distrust I feel towards the LLM itself and its inherent biases based on biases in the training data and the people that supervised this process
I think what they’re talking about is just for learning what a source might be, and then going and looking up that paper. In that sense it’s acting like just a very good search engine, and yes it saves you from having to look through millions of less relevant papers, but I don’t think it really dilutes much academically.
Meanwhile 90% of other video essayists are running all their scripts through chatgpt and it's so fucking obvious if you have the truesight of ever using these tools yourself, it drives me mad every time I go on youtube. And yet nobody notices, while Hank gets crucified for using it to… collate research sources? Moronic.
The amount of video essays I’ve watched on genuinely interesting topics that I’ve clicked off the moment they’ve said “its not x. its y” is sad
“Unhealthy AI use” holy shit yall are not living in reality. The entire thing is a non story. And the author of this article clearly didn’t do enough digging because the “pushback” line is already explained.
Using “I appreciate the pushback” once isn’t evidence of relying on AI. AI use such phrases often because people do, hence it’s in their training data a lot. If someone’s work is riddled with it, then yes it’s probably AI. But the anti-AI crowd seems to be on a witch-hunt.
I have a bridge to sell you if you really believe that crap.
He was reading the text for the show, and the “I appreciate the pushback” was there in the middle of it, without rhyme or reason.
I think you can read an implication from the dopamine hit comment that he is not using the chatbots merely for research. As you noted, a lot of people find themselves in an unhealthy, sometimes even physically dangerous, feedback loop with these tools -- which are, as yu note, designed to keep people engaged. Combined with his family's alarm, I suspect he is actually caught in one of those feedback loops, using the bots for far more than just research. It is certainly plausible, given the information in your post.
Emotional reliance for sureeeee
Every time I see a post like this, I discover a new AI pattern that I use naturally in human speech. I’d so fluidly say, “I appreciate the pushback.” It’s disturbing to me that so many people are reading human thoughts and presuming AI usage when it’s not always the case. Feels a little like a witch hunt. I totally agree that AI presents dangers, but I do wish we could find a better way of handling it than our current posture of suspicion, cancelling, and condescension. Not a critique of you—just a reflection on what’s happening overall.
> I’d so fluidly say, “I appreciate the pushback.”
To nobody there? While reading a script for the show? Where it doesn't follow from anything said previously?
Come on, this is beyond idiotic.
I ran your comment through an AI detector, and it said there’s a 99% chance you spent last Friday alone.
Yes, AI tools will say the most stupid shit - as will the people who outsource their thinking and writing to them.
Yep. My entire point. AI “detects” certain phrases and patterns because it was trained on actual human writing. The AI detection system is just as goofed up as the generation system. When the public is told that a certain pattern indicates AI, they categorically dismiss what context should clarify. The em dash has been villainized. If/then statements have been villainized. Now we can’t use the word “quiet” or the phrase “Thank you for the pushback.” We need thoughtful people writing about this with nuance, not lonely Twitter bros working out wounded egos with childish comments and bravado.
Maybe it’s because I work in the corporate world, but that’s a phrase I use and wouldn’t think anything of I heard someone else use it.
This whole essay borderlines on victim blaming, there's a reason AI psychosis is a thing and not everyone has the tools to resist what some of this LLMs offer. If the guy says he gets a "dopamine hit" from continuous AI use I mean... he's not alone, we need more regulations for AI and that's the bottom line
> This whole essay borderlines on victim blaming, there's a reason AI psychosis is a thing
He's no victim. He's making $$$ youtube ad money, while half-arsing his show's scripts to have them written by AI.
As shocking as it is to see people publish something created by LLMs, and obviously discussing ethical and romantic conundrums, I don’t see why people get so mad about research with LLMs? You want me to waste time watching and reading a bunch of “fashion advisors” pretending to be emotionally invested in my sartorial choices? Or can I just use a chatbot for a quick starting point and then progress according to my individual interests?
"His line about “dopamine” hits from LLMs is silly and dramatic."
Hank Green was diagnosed with ADHD as a child. I was diagnosed with ADHD as an adult. I can tell you from myriad personal experience that using AI is an excellent way to get a dopamine hit, especially when you're in a conversation with it about a hyperfocus that bores the pants off the actual humans in your life. (It's one of the reasons I enjoy using AI for brainstorming, etc. It makes writing much more fun.) Is that healthy? Sometimes yes, sometimes no, and it looks like Green hit his limit of healthy interaction. But it's not silly or dramatic for him to say that.
Like Bill Nye didn’t have a team of dozens of people helping him, funded by the government 🤣 give the guy a break! It’s not easy to compete online at his level. The amount of content these creators have to push out, with their own funding for a team, is crazy.
And yet you all use cameras that process your photos with AI and automation and algorithmic filters and use AI to give you headshots and more.
No, we don't "all".
And it's not the same, even for those that do. The picture of your plate on Instagram or your profile photo on some shit social site is not the same as an essay or a vlog.
My art is the product of forty plus years of study. You insult and you don’t even look. Not very bright are you? You’ve invalidate your own options straightaway. Is that the level of insight you bring to your writing?
> My art is the product of forty plus years of study.
Nope, you studied for 40 plus years, and now you decided to throw in the towel and phone it in with AI.
> You insult and you don’t even look.
As should be the case with all AI or AI-assisted art. Period.
You’re a bit of a daft twat or wanker, take your pick. My photography isn’t ai - i shoot with a Leica or an iPhone and I’ve never photographed a plate. Nor are my Antiselfies AI . If you’re getting this much wrong and being so judgemental and shouty how bad is any analysis in your writing? Highly shit I suspect. I certainly won't be looking at it. Get help.
I remember when The Fault in Our Stars came out. I was an eleven-year-old girl who curled up to read it under my peach polka dot duvet in the small box room that was meant to be a storage room. My wallpaper was duck egg green because my best friend's was duck egg green too. We posted our books on Instagram, back when it was vignette Pictochat filters, sleepover selfies, and showing off our new pumpkin mug with fairy lights in the background. We'd get 6 likes and we'd be happy. We'd talk about cigarettes being metaphors during recess, and then at home over Kik. We'd dip chicken nuggets in the vinegary sweetness of off-brand ketchup, read some more, and then go to sleep.
I remember my best friend with the duck egg wallpaper posting a photo of her crying at the cinema. Her and her other friends. They'd gone to watch the adaptation of pretty actors playing sick kids. They were holding blue slushies and wore chokers to try and look like cool teenage movie protagonists. We didn't know then that we were on the precipice of something dangerous--that those Pictochat filters would turn into air brushing, would turn into weight loss drugs that would kill some of the girls like us. We didn't know that, in ten years time, those 6 likes could be the end of a company, the focus of a life, the make-or-break of paying bills that month. We were just reading books next to fairy lights, drinking slushies and crying and going to sleep. We picked the wallpaper we wanted, not the one a YouTube manifestation guru told us to in order to "upgrade your chic". We trusted John Green to tell us things about mortality, empathy, and how people are people and not places. We felt and captured and thought about things. We spoke about them at recess, not just over Kik.
Soon, we'd be loosing the moment to a story photo--to a pre-meditated life plan, informed by robots and influencers and striking the perfect macros. I think we should go back to eating chicken nuggets for tea sometimes. I think we should put on some fairy lights, and curl up under a blanket, and face our mortality. We should use it a reminder to notice the duck egg wallpaper, and cry at movies, and to go to sleep at night, and not rely on men who forget what it is they once wrote, and sell away their empathy.
This is one of the most dramatic articles I’ve ever read lol.
I am so disappointed in Hank Green. He really should know better. Hadn't his channel been active for years before the AI boom? He should not have been having any problems researching and scripting videos.😭
I work for a massive global corp (unfortunately) and it’s funny reading that recording meetings and plugging them into Claude is shocking because of how normalised it is at work. Where I work now, senior staff have Claude agents that work / converse with other agents on their behalf when they’re out of office. Recording meetings instead of taking physical notes is honestly the very tip of the iceberg! One of our KPIs is to use AI for 60% of tasks at minimum.
Important caveat yes I’m looking at a career change
I'm sorry I'm sure you put a lot of effort into writing this but I could only think one thing while reading: who the fuck is Bill Nye?
for context: i am Italian and I am gen Z