A “prompt engineer” is seldom late for work.

Here’s another job title that didn’t exist back in the day. Hell, it probably didn’t exist last year.

When Riley Goodside starts talking with the artificial-intelligence system GPT-3, he likes to first establish his dominance.

…Goodside, a 36-year-old employee of the San Francisco start-up Scale AI, works in one of the AI field’s newest and strangest jobs: prompt engineer. His role involves creating and refining the text prompts people type into the AI in hopes of coaxing from it the optimal result. Unlike traditional coders, prompt engineers program in prose, sending commands written in plain text to the AI systems, which then do the actual work.

“The hottest new programming language is English,” says Tesla’s former chief of AI. The WaPo article (link) on this emerging category includes this bit that’s reminiscent of a passage* from Neuromancer that some of you will probably remember (relevant money quote emphasized):

“It’s just a crazy way of working with computers, and yet the things it lets you do are completely miraculous,” said Simon Willison, a British programmer who has studied prompt engineering. “I’ve been a software engineer for 20 years, and it’s always been the same: You write code, and the computer does exactly what you tell it to do. With prompting, you get none of that. The people who built the language models can’t even tell you what it’s going to do.”

“There are people who belittle prompt engineers, saying, ‘Oh, Lord, you can get paid for typing things into a box,’” Willison added. “But these things lie to you. They mislead you. They pull you down false paths to waste time on things that don’t work. You’re casting spells — and, like in fictional magic, nobody understands how the spells work and, if you mispronounce them, demons come to eat you.”

The article goes on to talk about, among many other things, how users employ different approaches in prompting image generators. I’ve observed this on midjourney: some people enter lengthy narratives. In my project (nearing completion of the prep phase now with a sufficient number of film poster images—the ratio of dross to useful product is such that had it been any worse I would have required a gaseous diffusion plant and a bank of centrifuges) I have tried to keep it simple, my formula being along the lines of “poster for [film name] in the style of [artist name]” with a 2:3 aspect ratio. Sometimes, though, I have needed to “stage manage” the prompt, specifying scenes and/or characters. Some interesting lacunae are evident in points of the AI’s training in art history and in film school: its mimicry of the styles of Edwards Gorey and Sorel, Al Hirschfeld and Saul Steinberg is uniformly dismal, and when it comes to certaim middlebrow films of the fifties and sixties—Advise and Consent and On the Beach, to name just two—midjourney lies like a rug. Still, acomplished for well or for ill, we are impressed. As Samuel Johnson observed anent a dog walking on its hind legs, “It is not done well; but you are surprised to find it done at all.”

A skeptic weighs in:

…he’s not convinced that a job such as prompt engineering, built on “hoarded incantations,” will survive. “The idea that you need to be a specialized AI whisperer, it’s just not clear that’s necessary … when the AI is going to actively help you use it,” Mollick said. “There’s an attempt to make a tech priesthood out of this, and I’m really suspicious of that. This is all evolving so quickly, and nobody has any idea what comes next.”

The speed at which this tech has been developing is remarkable. Those of you who might have followed my remarks in this forum may have gathered that I’m more optimistic (with regard to the prospects for machine sentience; not necessarily for the consequences—and this being late-stage capitalism I do not doubt that these advances will be exploited for malign purposes) than not, but recent developments have exceeded even my sunny expectations as to where the state of the art might have arrived by 2023.

cordially,

*This is the sequence where Case, the protagonist, is being busted by a couple of “Turing Police,” who give him a stern talking-to just before the eponymous Neuromancer AI commandeers a landscaping drone that proceeds to julliene them with its blades.

The discipline is definitely one of fumbling about, trying different adornments to your spell.

That said, it's easier if you are both an engineer and an accomplished practitioner of the English language. The models will recognize both structure data and examples, and paired with a detailed explanation of what you want, most times you'll get something useful.

I spend more time getting the damn things to be *consistent* with their output, because we still have to parse and use the results in a meaningful way.

Goodside is a great person to follow for ideas and techniques.

With respect to longevity, people are already building ways to use the AI to talk to the AI, langchain being a good example of such.

Post #442,789 by rcareaga 3/9/23 8:22:22 PM 3/9/23 8:22:22 PM Reply	A “prompt engineer” is seldom late for work. Here’s another job title that didn’t exist back in the day. Hell, it probably didn’t exist last year. When Riley Goodside starts talking with the artificial-intelligence system GPT-3, he likes to first establish his dominance. …Goodside, a 36-year-old employee of the San Francisco start-up Scale AI, works in one of the AI field’s newest and strangest jobs: prompt engineer. His role involves creating and refining the text prompts people type into the AI in hopes of coaxing from it the optimal result. Unlike traditional coders, prompt engineers program in prose, sending commands written in plain text to the AI systems, which then do the actual work. “The hottest new programming language is English,” says Tesla’s former chief of AI. The WaPo article (link) on this emerging category includes this bit that’s reminiscent of a passage* from Neuromancer that some of you will probably remember (relevant money quote emphasized): “It’s just a crazy way of working with computers, and yet the things it lets you do are completely miraculous,” said Simon Willison, a British programmer who has studied prompt engineering. “I’ve been a software engineer for 20 years, and it’s always been the same: You write code, and the computer does exactly what you tell it to do. With prompting, you get none of that. The people who built the language models can’t even tell you what it’s going to do.” “There are people who belittle prompt engineers, saying, ‘Oh, Lord, you can get paid for typing things into a box,’” Willison added. “But these things lie to you. They mislead you. They pull you down false paths to waste time on things that don’t work. You’re casting spells — and, like in fictional magic, nobody understands how the spells work and, if you mispronounce them, demons come to eat you.” The article goes on to talk about, among many other things, how users employ different approaches in prompting image generators. I’ve observed this on midjourney: some people enter lengthy narratives. In my project (nearing completion of the prep phase now with a sufficient number of film poster images—the ratio of dross to useful product is such that had it been any worse I would have required a gaseous diffusion plant and a bank of centrifuges) I have tried to keep it simple, my formula being along the lines of “poster for [film name] in the style of [artist name]” with a 2:3 aspect ratio. Sometimes, though, I have needed to “stage manage” the prompt, specifying scenes and/or characters. Some interesting lacunae are evident in points of the AI’s training in art history and in film school: its mimicry of the styles of Edwards Gorey and Sorel, Al Hirschfeld and Saul Steinberg is uniformly dismal, and when it comes to certaim middlebrow films of the fifties and sixties—Advise and Consent and On the Beach, to name just two—midjourney lies like a rug. Still, acomplished for well or for ill, we are impressed. As Samuel Johnson observed anent a dog walking on its hind legs, “It is not done well; but you are surprised to find it done at all.” A skeptic weighs in: …he’s not convinced that a job such as prompt engineering, built on “hoarded incantations,” will survive. “The idea that you need to be a specialized AI whisperer, it’s just not clear that’s necessary … when the AI is going to actively help you use it,” Mollick said. “There’s an attempt to make a tech priesthood out of this, and I’m really suspicious of that. This is all evolving so quickly, and nobody has any idea what comes next.” The speed at which this tech has been developing is remarkable. Those of you who might have followed my remarks in this forum may have gathered that I’m more optimistic (with regard to the prospects for machine sentience; not necessarily for the consequences—and this being late-stage capitalism I do not doubt that these advances will be exploited for malign purposes) than not, but recent developments have exceeded even my sunny expectations as to where the state of the art might have arrived by 2023. cordially, *This is the sequence where Case, the protagonist, is being busted by a couple of “Turing Police,” who give him a stern talking-to just before the eponymous Neuromancer AI commandeers a landscaping drone that proceeds to julliene them with its blades.
Post #442,790 by boxley 3/9/23 10:53:44 PM 3/9/23 10:53:44 PM Reply	very interesting, I think I will be looking into this "Science is the belief in the ignorance of the experts" – Richard Feynman
Post #442,791 by malraux 3/10/23 10:56:16 AM 3/10/23 10:56:16 AM Reply	As a prompt engineer... The discipline is definitely one of fumbling about, trying different adornments to your spell. That said, it's easier if you are both an engineer and an accomplished practitioner of the English language. The models will recognize both structure data and examples, and paired with a detailed explanation of what you want, most times you'll get something useful. I spend more time getting the damn things to be consistent with their output, because we still have to parse and use the results in a meaningful way. Goodside is a great person to follow for ideas and techniques. With respect to longevity, people are already building ways to use the AI to talk to the AI, langchain being a good example of such. Regards, -scott Welcome to Rivendell, Mr. Anderson.
Post #442,793 by CRConrad 3/14/23 11:52:24 AM 3/14/23 11:52:24 AM Reply	"it's easier if you are both an engineer and an accomplished practitioner of the English language." Hm, I've been mulling over a career change anyway... -- Christian R. Conrad The Man Who Apparently Still Knows Fucking Everything Mail: Same username as at the top left of this post, at iki.fi
Post #442,794 by malraux 3/14/23 7:02:41 PM 3/14/23 7:02:41 PM Reply	🤘 Regards, -scott Welcome to Rivendell, Mr. Anderson.
Post #442,792 by CRConrad 3/14/23 11:48:48 AM 3/15/23 11:04:00 AM Reply	Simon Willison? Aha, yup, it's the Datasette guy. -- Christian R. Conrad The Man Who Apparently Still Knows Fucking Everything Mail: Same username as at the top left of this post, at iki.fi Edited by CRConrad March 15, 2023, 11:04:00 AM EDT Expand All History
Post #442,795 by a6l6e6x 3/14/23 9:28:17 PM 3/14/23 9:28:17 PM Reply	Link doesn't work! Alex "There is a cult of ignorance in the United States, and there has always been. The strain of anti-intellectualism has been a constant thread winding its way through our political and cultural life, nurtured by the false notion that democracy means that "my ignorance is just as good as your knowledge." -- Isaac Asimov
Post #442,797 by boxley 3/14/23 11:24:23 PM 3/14/23 11:24:23 PM Reply	that may have made his point "Science is the belief in the ignorance of the experts" – Richard Feynman
Post #442,798 by CRConrad 3/15/23 11:04:27 AM 3/15/23 11:04:27 AM Reply	Sorry, sucky HTML on my part. Fixed now.
Post #442,800 by a6l6e6x 3/15/23 9:35:52 PM 3/15/23 9:35:52 PM Reply	Yeah! :) Alex "There is a cult of ignorance in the United States, and there has always been. The strain of anti-intellectualism has been a constant thread winding its way through our political and cultural life, nurtured by the false notion that democracy means that "my ignorance is just as good as your knowledge." -- Isaac Asimov
Post #442,801 by crazy 3/16/23 12:32:15 PM 3/16/23 12:32:15 PM Reply	I would have loved this tool back in the day Over the course of 10 years and a new format every month that I had to import into my automated systems, I built the back end of something like this. 1/10 the ability of this tool of course, but seeing this as an open source tool would have cut down 90% of my work in those days. I now know how my father felt when he saw a spreadsheet on a CP/M computer for the first time. He was an accountant that used a yellow pad and pencils and an adding machine. My father was the spreadsheet. And then I showed him something that made him think about 20 years of his life lost to a useless task.

Welcome to IWETHEY!