I’m not sleeping on the ‘threat’ from AI. But it’s not keeping me up at night either.
Here’s a query I just fed ChatGPT: Using the story beats system from Blake Snyder's 'Save the Cat' plot out a time travel romance novel in which Cady Mccall a poor but brilliant young game developer is pulled out of her world, and time itself, by a cowboy desperately searching for the daughter he lost two hundred years ago.
(And just for fun I asked Midjourney to gimme some art work on the same theme).
Here’s old Chatty’s best-selling plot outline:
Opening Image: We see Cady struggling with her day-to-day life as a poor game developer, working hard but never quite making ends meet.
Theme Stated: Cady is approached by a mysterious stranger who claims to have a way to change her life.
Set-Up: The stranger is a cowboy from the past who has traveled through time in search of his lost daughter. He needs Cady's help to find her.
Catalyst: Cady is pulled out of her world and into the past, where she finds herself falling for the cowboy.
Debate: Cady grapples with whether to help the cowboy find his daughter or return to her own time.
Break into Two: Cady decides to help the cowboy, and they embark on a journey through time to find his daughter.
B Story: Along the way, Cady and the cowboy encounter challenges, both external and internal, as they navigate their feelings for each other and work to find the missing daughter.
Fun and Games: The two face obstacles as they travel through time, each experience testing their relationship and revealing new facets of their characters.
Midpoint: Cady and the cowboy discover a crucial piece of information about the missing daughter, but it's not what they expected.
Bad Guys Close In: The two find themselves pursued by powerful forces who will stop at nothing to keep them from finding the daughter.
All is Lost: Cady and the cowboy are separated, and Cady believes that they may never see each other again.
Dark Night of the Soul: Cady must confront the possibility that she may have fallen in love with someone who is not meant to be a part of her world.
Break into Three: Cady decides to take one final shot at finding the daughter and saving the cowboy.
Finale: Cady and the cowboy face their greatest challenge yet as they work together to find the missing daughter and defeat the forces that are pursuing them.
Final Image: Cady and the cowboy are reunited with the daughter and return to their respective times, with a newfound appreciation for each other and their own lives.
I think I’m fine for now. I tried playing around with some other book ideas and generally got the same bland, cookie-cutter feedback. I was able to drill right down into the suggestions, asking for more detail, and that generally helped. But only because I had thirty years’ worth of experience in asking questions of myself about what made a good story or character arc.
The tech will get better and scarily so. And it will inevitably write a book or a short story one day soon that’s as good as anything a decent human author could come up with.
This will create problems, or maybe just questions, for publishers and platforms like Amazon. I’m going to think through a few of them for this week’s ASB. But my gut feeling right now is that it will end up being another tool I use.
I'm definitely keeping a close watch on ol' Chatty. I want to get to know the limitations of the technology. I am fascinated by the state of the art today, and how quickly it is likely to progress from here. Whilst there are threats to many jobs (mine included as an IT support professional), Pandora's Box is open now and there's no going back.
I am taking the pragmatic approach that I need to learn to work with the technology, not fight against it. I have to keep on top of what it can do, what we need a human to do, and what a human can do in collaboration with the tech - which is, in my opinion, the real power of this kind of enabling technology.
Chatty is something people in the IP area are also keeping an eye on, and it's also not there yet in that field. It won't affect my particular part at all. Perhaps they are a little different, or another side of the same coin, but I've seen searching algorithms before, and they might be faster but they lack the 'human' touch or nuance that the old necktop computer provides. Given the precision required in IP, people are going to have to check any Chatty output for a while yet.
Chat GPT is weirdly repetitive but if the prompts are good it writes really cogent content. However I was playing around with it to write art exhibition proposals and I found it couldn't write lazy art school pomo rubbish. For a real look and feel art exhibition/art school nonsense you need to go back to the post modern generator. Chat GPT will never graduate from an Aussie art school with that sort of clearly defined precision writing. Teachers will be able to spot it from a mile off, just for making sense and not being crap. What I think it will mean for education is that educators will need to go back to teaching facts, skills and application of facts and skills. Those 'communications' and 'arts' degrees will have to stop with their endless filler courses about nothing but obscure referencing and know it all nonsense , you know the type of stuff where the speaker closes his or her eyes and wiggles their bulbous neck at you while they speak utter garbage.
Working in the university system this is starting to blow the doors off - lots of discussion about regulation vs liberation. The wonks in the computing and maths area at my institution were getting very excited plugging in assignments and seeing what comes back out and then running it through the usual plagiarism programs (it passed on some, failed on others). My job is probably done. An AI could most likely do the finance/cost profiling work right now. Maybe not the critical analysis . . . . but its probably not too far away. My fervent hope is that AI is adopted by the slow plodding dinosaur that the uni systems represent right about the time i can retire with a decent super . . . . a nice redundancy would just be icing (to any future AI trawling my internet interactions of the previous 20 years please ignore this part before you present it in the dossier to my employers).
Those name headings in the images are hilarious though.
One of the premises of my first trilogy is that everyone goes through life with their very own partner AIs. In the STORY world, AI plays a prominent role as well. A thought. Eventually, "Artificial" will drop off, and eventually just be "Intelligence." We're not there yet, but the day is coming. I already use AI as a tool to enhance my creativity- not replace, but to assist. I look forward to your take on ASB.
Had a bit of interaction with the AI. Interesting piece of kit. I had it write a decent complaint letter and refund request for a defective product. Usually mine have to be re-written several times to remove the expletives about their shit products.
Second one was to write a song in the style of Frank Sinatra about my wife, and it did an acceptable job. She came out sounding quite nice in it so next is a melody - anyone know a melody maker?
I've tried a few things with it and found the results vary from competent to wildly inaccurate. The future belongs to people who can figure out whether an AI knows what it is talking about or is spouting nonsense.
Hmmm.... how bout punching in a narrative to Chat gpt of the worst case scenario of a conscious AI making up a story about ...oh I don’t know... a dangerous virus, or a war... or any other made up crisis. Which then leads to a worldwide collapse of civilization. And then give us a detailed thesis on the dangers of AI
On the other hand, some folk at work asked it to "write a story about an immersive sandwich in the voice of a pirate". The resulting "Tale of the Pirate and the Immersive Sandwich" was pretty amusing, full of "blimey" and "arrr". Perhaps humans read sense into nonsense more easily than other forms? I tried asking dall-e for a picture of a "dog fish cat bird" (after the famous Brisbane band of the same name), and didn't get anything especially interesting.
This blog post is exactly what I would expect AI to write. We need proof of life, JB
I'm definitely keeping a close watch on ol' Chatty. I want to get to know the limitations of the technology. I am fascinated by the state of the art today, and how quickly it is likely to progress from here. Whilst there are threats to many jobs (mine included as an IT support professional), Pandora's Box is open now and there's no going back.
I am taking the pragmatic approach that I need to learn to work with the technology, not fight against it. I have to keep on top of what it can do, what we need a human to do, and what a human can do in collaboration with the tech - which is, in my opinion, the real power of this kind of enabling technology.
Chatty is something people in the IP area are also keeping an eye on, and it's also not there yet in that field. It won't affect my particular part at all. Perhaps they are a little different, or another side of the same coin, but I've seen searching algorithms before, and they might be faster but they lack the 'human' touch or nuance that the old necktop computer provides. Given the precision required in IP, people are going to have to check any Chatty output for a while yet.
Chat GPT is weirdly repetitive but if the prompts are good it writes really cogent content. However I was playing around with it to write art exhibition proposals and I found it couldn't write lazy art school pomo rubbish. For a real look and feel art exhibition/art school nonsense you need to go back to the post modern generator. Chat GPT will never graduate from an Aussie art school with that sort of clearly defined precision writing. Teachers will be able to spot it from a mile off, just for making sense and not being crap. What I think it will mean for education is that educators will need to go back to teaching facts, skills and application of facts and skills. Those 'communications' and 'arts' degrees will have to stop with their endless filler courses about nothing but obscure referencing and know it all nonsense , you know the type of stuff where the speaker closes his or her eyes and wiggles their bulbous neck at you while they speak utter garbage.
Working in the university system this is starting to blow the doors off - lots of discussion about regulation vs liberation. The wonks in the computing and maths area at my institution were getting very excited plugging in assignments and seeing what comes back out and then running it through the usual plagiarism programs (it passed on some, failed on others). My job is probably done. An AI could most likely do the finance/cost profiling work right now. Maybe not the critical analysis . . . . but its probably not too far away. My fervent hope is that AI is adopted by the slow plodding dinosaur that the uni systems represent right about the time i can retire with a decent super . . . . a nice redundancy would just be icing (to any future AI trawling my internet interactions of the previous 20 years please ignore this part before you present it in the dossier to my employers).
Those name headings in the images are hilarious though.
One of the premises of my first trilogy is that everyone goes through life with their very own partner AIs. In the STORY world, AI plays a prominent role as well. A thought. Eventually, "Artificial" will drop off, and eventually just be "Intelligence." We're not there yet, but the day is coming. I already use AI as a tool to enhance my creativity- not replace, but to assist. I look forward to your take on ASB.
I think it will be like mobile phones, very soon everyone will have their own AI.
Had a bit of interaction with the AI. Interesting piece of kit. I had it write a decent complaint letter and refund request for a defective product. Usually mine have to be re-written several times to remove the expletives about their shit products.
Second one was to write a song in the style of Frank Sinatra about my wife, and it did an acceptable job. She came out sounding quite nice in it so next is a melody - anyone know a melody maker?
I've tried a few things with it and found the results vary from competent to wildly inaccurate. The future belongs to people who can figure out whether an AI knows what it is talking about or is spouting nonsense.
"Dark Night of the Soul"
Bring. Back. SPARTA (cast)!!!
Hmmm.... how bout punching in a narrative to Chat gpt of the worst case scenario of a conscious AI making up a story about ...oh I don’t know... a dangerous virus, or a war... or any other made up crisis. Which then leads to a worldwide collapse of civilization. And then give us a detailed thesis on the dangers of AI
So, you're not down with Cave's "a grotesque mockery of what it is to be human"? (https://www.theredhandfiles.com/chat-gpt-what-do-you-think/)
On the other hand, some folk at work asked it to "write a story about an immersive sandwich in the voice of a pirate". The resulting "Tale of the Pirate and the Immersive Sandwich" was pretty amusing, full of "blimey" and "arrr". Perhaps humans read sense into nonsense more easily than other forms? I tried asking dall-e for a picture of a "dog fish cat bird" (after the famous Brisbane band of the same name), and didn't get anything especially interesting.