Generative Artificial Intelligence (GenAI) is not the AI for me. Machine Learning, for example the type that uses Bayesian algorithms, are totally different.
Business use of GenAI
I spoke with a developer recently who has been tasked with using AI more and more in his day-to-day. (why is a good question that I’ll get to)
He told the AI agent to write a piece of code that would have taken him ten seconds to write. When he reviewed the output, he determined that yes, this code works. He also discovered that it cost the company $0.28 worth of usage tokens
28 cents for ten seconds of work.
What if this thing was coding full-time? a ten second block happens six times per minute.
0.28*6=1.68 so that’s $1.68 per minute
There’s 60 minutes in an hour, so the company would be paying this chatbot $100.80/hour
40 hours in a work week means $4032 hours a week
52 work weeks in a year, so that’s $209,664 a year
That is more than that particular developer’s salary.
One can look at all of the information available about GenAI currently deployed in production environments in websites, data centers, manufacturing centers, warehouses, logistics, transportation, etc., and what many of them have in common is that
- Real human developers need to write the prompts (time=money)
- Real human developers need to verify what the AI produces (time=money)
- Real human developers need fix what it produced and/or re-submit with new prompts (time=money)
- Real humans need to perform quality assureance before pushing to production(time=money)
- Real human developers need fix everything that it broke once it gets pushed to production(time=money)
That’s a lot of time and money; a lot more than if businesses just paid human developers.
So not only is AI plagiarizing my writing, plagiarizing my artwork, destroying the environment (documented and demonstrably cases of people becoming sick or dying from AI data centers), and facilitating mass illegal and unconstitutional mass surveillance of citizens around the globe (I can confirm the Five Eyes nations (ustralia, Canada, New Zealand, the United Kingdom, and the United States), the People’s Republic of China, and The Russian Federation), and giving regular users and business users wrong information (“Pregnant women can eat up to five cigarettes a day”) and performing harmful actions (Amazon Prompt “Engineer”: reduce the current order queue load in the most efficient way possible. Amazon’s internal use AI: <deletes entire Amazon order queue>) – but it’s actually costing you more, as a business, to use it, than to not use it!
So why is it being pushed at any given company? Well it’s making the companies that own these things lots of money, and, it is training the AI models to get even better to make those techbros even more money. You can do the math and connect the dots yourself.
Personal use of AI
I’m an artist. I’m a photographer. I’m a writer. I make traditional art and digital art. I’ve done some voice acting. I’ve done some on screen acting. I’m a voice actor. I’m an actor. No one can take that away from me, because that’s what I am. I always have been, and I always will be, even after my death.
In fact, all of the above are types of artistic expression, so anyone with any one of those things, if I don’t wish to be overly verbose, are art, made by me, an artist.
Some of the people that I’ve spoke to about their own personal use of GenAI to produce what they call “art”. All of them have told me “I can’t draw.” or “I can’t write.” or “I can’t take photos.” or “I can’t use the software for making digital art.”, and whenever any of them have acquired to my request to let me see something they wrote or drew or photographed or worked on in Blender or Illustrator or Photoshop – despite them telling me that it is garbage or that it sucks – I have always had the same reaction.
“Wow. That person made this. They’re a real artist, just like me.”
All real artists are going to have that reaction as well. People that see what they created and ridicule aren’t artists, they’re haters. And don’t get me wrong, sometimes haters can draw a really really good picture, or paint an amazing painting, etc., but they’re still haters. Don’t listen to them. And that ridicule isn’t a reason to use GenAI instead of making your own stuff because guess what? Other GenAI users are going to ridicule whatever you instructed GenAI to produce, not matter what it looks like. Haters are gonna hate.
“Wow. That person made this. They’re a real artist, just like me.” I reiterate. That’s what real artists say when they see art. It does not matter how “good” or “bad” anyone subjectively thinks that it is. What’s objectively true is that it was made by a person producing art, ergo it was made by an artist, thus it is art.
GenAI does not produce art.
When I see someone show something off that they prompted GenAI to produce, and call that art… that is the time I laugh. But I also feel pity for them because the sphere of influence that they live in – society as experienced by them – have conditioned them to believe that they are less-than, and that their art is unworthy. I also feel anger on their behalf against that society, as well as anger towards them because of all of the above.
To quote an animated genius narcissistic alcoholic, “Shitting the bed is not better than not shitting the bed.” By this I mean “You didn’t have to do that. You didn’t have to use GenAI. You did not have to contribute to the plagiarism, pollution, poverty, death, surveillance/control, illness, (and mostly outside of this nation) indentured servitude and child slavery, war, and (globally) the transformation of the planet into one in which you – the average person reading this – will not survive.
Is all AI bad?
Machine learning is technically AI.
It’s the kind of thing that is not producing media like text, images, audio, video. It’s analyzing data. It’s using mathematics and algorithms to make conclusions. It is (mostly) not using massive data centers, or any data centers.
It’s filtering your email for spam. It’s detecting cancer cells and finding cures to cancer. It’s doing that kind of stuff.
It’s even making your video games more fun!
In 1996, Star Wars: Shadows of the Empire was released for the Nintendo 64. An almost always overlooked fact about the game is that it did incorporate machine learning at runtime. It’s very telling that it is even more difficult to find information on this now than it was in the 90s and early 2000s.
Here’s what I mean: The game had an algorithm in it that would record where the player walked during the walking levels of the game. The player-character is walking around the Rebel Alliance base on Hoth as it’s being invaded by the Galactic Empire. You, the player, encounter some storm troopers. They overwhelm you and kill your character. You start over from your last check point, but now you have an advantage because you know where the game has placed those storm troopers. There’s still too many of them, and you have to start over again. Then you get past that part, but get taken out in the next hallway by a lone storm trooper that took advantage of that fact that you were already half dead from the last firefight. You try again, this time making it through the first group really easily, take out the guy in the next hall, but then are mowed down by two more enemies in the next room. You try it again, but this time it’s different. This time, the game’s algorithm has determined where you are going to go because you’ve been going that way over and over, and so it moves the storm troopers out of your path – out of harms way – and into a position where they easily flank you and eliminate you. The devs thought that would keep people engaged and entertained because knowing where the bad guys are is boring. I agree with them. Maybe you do, maybe you don’t, but the point is this:
That “A.I.” did not require an Internet connection to connect to a data center to steal your moves or surveillance center to build a secret profile about you. That “A.I.” did not utilize a harmful data center. That “A.I.” did not rely on plagiarism. That “A.I.” did not facilitate plagiarism. That was all contained within a little game cartridge, and lived inside the compute of your game console that was connected to nothing but your electrical outlet and your television.
There’s also a lot of AI use in modern CCTV. Many of those do connect to a cloud to do all the bad A.I. things. Many brands do not. My favorite cameras are the ones that perform the computing themselves. These embedded-and-deployed devices are called “edge devices”, and so what they are doing is called “edge compute”. The tiny little camera that fits in the palm of my hand looks at contrast in the image to detect an object and decide if it’s shaped like a box, or a dog, or a person, or a car. And then sends me a notification. “Oh, animal detected? I’m not going to interrupt what I’m doing to look at a squirrel right now. Maybe later.” versus “Oh package detected? I had better go grab it before a porch pirate.” or “Person detected? I’ll take out my phone and talk to them through the intercom to see what they want.”