I'm excited for this portion to be done -- but it's going to be the easy portion. The rest is going to be more of a challenge.
I do like the Neil Armstrongesque footprint in the sand in the second photo.
1. What does it mean, exactly, if one is "asasdf at kajdsf"?
2. Why are there two people on LinkedIn who fill this particular role?
3. Why do their profiles come up as 404 errors?
Could it be . . . nah. People wouldn't lie about being asasdf at kajdsf, would they?
The text (available here in context, with comments):
Posting this on behalf of a member who would like to remain anonymous:
"I’m an art director and supervisor for a large studio. The studio heads had the bright idea before I started to hire prompters. Several bros were brought onto the film project. I absolutely hated myself for not quitting on the spot but stuck with it because it’s mercenary out there. Have a family to feed etc. I decided to use this time wisely. Treat them as I would any artist I had hired. First round of pictures of a sweeping Ariel forest landscape comes through and it’s not bad. They submit a ton of work and one or two of the 40 are ok. Nearly on brief. So first round feedback goes through and I tell them about the perspective mistakes, colour changes I want, layers that any matte painting would be split into. Within a day I get 5 variants. Not changes to the ones I wanted but variations. Again. Benefit of the doubt I give them another round of feedback making it clear. Next day it’s worse. I sit there and patiently paint over, even explaining the steps I would take as a painter. They don’t do it, anomalies start appearing when I say I want to keep the exact image but with changes. They can’t. They simply don’t have the eye to see the basic mistakes so the Ai starts to over compensate. We get people starting to appear in the images. These are obviously holiday snaps.
“Remove the people”
“What would you like them changed to?”
“… grass. I just don’t want them there”
They can’t do it. The one that can actually use photoshop hasn’t developed the eye to see his mistakes, ends up getting angry at me for not understanding he can’t make specific changes. The girl whose background was a little photography has given me 40 progressively worse images with wilder mistakes every time. This is 4 days into the project.
I’m both pissed about the waste, but elated seeing ai fall at the first hurdle. It’s not even that the images are unusable, the people making them have no eye for what’s wrong, no thicker skin for constructive criticism and feedback, no basic artistic training in perspective and functionality in what they’re making.
Yes the hype is going to pump more money into this. They won’t go anywhere for a while. But this has been such a glowing perfect moment of watching the fundamental part fail in the face of the most simple tasks. All were fired and the company no longer accepts Ai prompters as applicants. Your training as an artist will always be the most important part of this process and it is invaluable. I hope this post gives you a boost in a dark time."
This harks back, unsurprisingly, to what we're seeing with AI and the written word, or indeed with anything: Those using AI are taking shortcuts and aren't developing the skills they could be using to create on their own without automated assistance. Yes, AI is a heck of a lot faster, but there's no learning. There's no investment, outside of knowing how to tweak what one asks the AI to do.
Being a naturally distrustful person, I read the story first, then went back to see what AI would have to say. I asked the second question: HOw many students have worked at the station?
Meta AI responded:
(I apologize for the potato-like quality of these screencaps.)
So it made a guess. It couldn't tell me how many precisely, because of course the story didn't offer a number. So it looked at what numbers it had and came up with the answer "many hundreds."
I guess a human could have come up with that answer. So why did I need AI for that? I don't know.
I still don't like the shortcut-edness of it all. Reading the story myself took only a few minutes; having AI do the reading and give me a response wasn't a timesaver, particularly in light of the noncommittal answer. Maybe with something more complicated it would be more helpful. But I'd still want to do the reading to make sure the AI wasn't misinterpreting something.
But what if I ended up misinterpreting something myself? Always a possibility. Still, it's worth checking myself.
I want to write about artificial intelligence, or, more specifically, about large language models such as ChatGPT and the like. But first, I’m going to write about Disney animator Milt Kahl. The reasons for this, I hope, will be clear.
Affectionately known as one of Disney’s “Nine Old Men,” Kahl is considered the animator’s animator, known for his skill in drawing fluid two-dimensional characters of the like of Shere Khan from The Jungle Book, Little John from Robin Hood, and Tigger from Disney’s Winnie the Pooh stories from the 1960s and 1970s.
He developed signature animations without the use of human models because he wanted to animate how the unique drawn figure would move, not as a model would imagine them moving. He’s known for the Kahl “head swaggle,” seen in the likes of Shere Khan and Edgar, the scheming butler from The Aristocats, to show off his ability to maintain head shape and body weight and positioning throughout the complex animation.
YouTuber A Humble Professor, an admirer of “animation, film, and comics,” analyzes scenes from 1977’s “The Rescuers,” for which Kahl was the directing animator. The professor focuses on animation of Madame Medusa, the film’s primary protagonist: a vain, vile woman of seemingly humble means in New York City who longs for her dreams of wealth to be realized through the discovery of a diamond long-lost in pirate treasure. The professor discusses rough animation of one scene: “Despite how rough these are, Medusa’s character shines right through them. We can really feel her frustration she feels in the scene and we can also see how her pear-shaped body influences how she sits down and scoots the stool.”
In other words, while Kahl could have used human models, or relied on past animation success (which he did with the head swaggle) he also realized that good character animation relied on how each unique character moved and talked and walked and, in the case of Madame Medusa, shrieked and climbed up on a stool and gathered her skirts when Bernard, a hero mouse in the film, arrived, spat out by one of her pet alligators after she whopped it on the head with her walking stick.
Kahl helped Medusa become a standout character because, through long effort and practice, he figured out how she looked and walked.
Uniquely.
Now let’s get back to Chat GPT and the like.
Essayist Evan Puschak, narrating a video essay called “The Real Danger of ChatGPT” at Nerdwriter1, says “Language is how human beings understand themselves and the world. But writing is how we understand uniquely. Not to write is to live according to the language of others, or worse, to live through edits, tweaks, and embellishments to language generated by an overconfident AI chatbot.”
Let me re-emphasize what Puschak says: Writing is how we understand uniquely. Not through the language of others, but through our own language and understanding.
That, I believe, is what large language models threaten to remove from us. Artificial intelligence will win out not because it develops the ability to think like humans, but because it will entice humanity with the expeditiousness that only results in humans accepting writing like an algorithm thinks they should is good enough.
Large language models present the world of writing with its own calculator moment and we as writers, educators, and students have to figure out how much we’re willing to give up for the ease of what this new calculator offers.
This is not to say that the likes of ChatGPT are bad in every way. Like any other tool, large language models can and should have their uses. The danger lies in confronting every writing problem as a nail, and using large language models as the ubiquitous hammer.
For every student and teacher who recognizes ChatGPT is a valuable tool for outlining, for brainstorming and the like, there is a student who takes the large language model shortcut and submits artificially-written work and fails classes because of it. Just this year I saw a student who was writing passable essays and commentary as an English learner succumb to the promise of ChatGPT – once, then again on a major assignment after being warned about the first offence; a semester of effort turn into failure in only a matter of days. My colleagues all share similar tales of woe. I can often hear my wife, who also teaches English, ranting in the next room in our basement as she finds another student who thought artificial intelligence was their best writing friend.
And I feel like a failure to him: Should I have done more to warn my students about the pitfalls of AI? Did I miss AI on previous assignments and not nip the temptation in the bud earlier, so his confidence built in the tool he was using? Is the course itself – its design out of my hands and in those of a committee trying their best – flawed? Is my teaching too lackadaisical?
I don’t know. But it’s hard to place all the blame on the student when the solution to a writing problem seems so clear and easily accessible – and as a savings of time.
How do I help my students hone their ability to understand uniquely the world they live in, rather than – if you can bear another metaphor – exchange it all for a mess of pottage?
Or should I, as some commenters on large language model-adjacent videos I’ve watched, just surrender because AI is only going to get better and become more human and more undetectable?
I’ve piled on the questions here. Let’s go find some answers.
Got a message tonight on LinkedIn from Justin Hegyi from the fine folks at The Outlier Team, inviting me to joing their crew of writers training large language models and artificial intelligence to write more gooder.
They singled me out, they say, because I am "both proficient in English . . . and are fluent or native Italian spakers -- like you!"
Italian must be the AI misspelling of French, because while I am somewhat proficient in French, my Italian is basically what I learned from A Christmas Story.
Thing is, this same guy contacte me back in February with the same message -- but he was working for a different company. To me, that's not necessarily a good look if you're looking to acquire talent.
My feelings on all of this, and AI:
I keep getting tagged on LinkedIn by companies that want to hire me to train AI to write better.
Then I see my students turning to AI to write their essays and gazing in wonder at the absolute nonsense AI produces and I conclude:
1. Yes, AI *does* need a lot of someones to teach it how to write.
2. Those who want to teach it how to write ought to be [word that will be censored by Facebook].
As much as I like Martin Short as a performer, this book was a little tedious to get through.
Reading about celebrities at parties is . . . dull. And it feels like the book is about one-third that. Maybe it's because I don't like parties myself.
Getting a glimpse of his life, that was interesting. Hearing from some of his characters, that was fun. But I'm gonna conclude by saying that watching Martin Short is a lot more entertaining than reading about him.
