I don’t often talk about computer technology and software here because A) it’s generally pretty boring, and B) there are far better sources for information about this stuff than me. But I want to talk about “artificial intelligence”, or AI as it’s known, because it has some serious implications for all of us as it comes into more widespread use. While I used to be a programmer I’m not so much interested in the technology of AIs as I am in the ethical and social implications of AIs, and that’s what I want to talk about briefly.
Take a look at that painting up there. It isn’t bad, really. Personally I’d like it if it were a bit brighter, but overall it isn’t horrible. It’s sort of a generic winter scene, peaceful, pleasant even.
Or what about this one —
Both of these paintings are originals that I just did in the last few days, but I didn’t make them in the usual way by applying paint to a surface or even a using a stylus on a drawing pad as I usually do. These were done by an app on my cell phone that is available for free on Apple’s app store called Draw Things. And as you can see it is pretty damn good. Well, sometimes. I will admit that about 8 out of 10 of the images it generates are, frankly, a bit, well, surreal. Like this one below.
But that being said, that it works at all is, I think, a bit amazing. What’s even more amazing is that unlike some of the other art AIs out there which actually run on some massive server somewhere out there on the internet, Draw Things runs entirely on the phone, no external processing power required. You can try it yourself although I’ll warn you that your phone will need a lot of free memory (the libraries the software uses are a gigabyte+ in size and you need at least one for the program to work) and using the program frequently will suck up a hell of a lot of battery as well because it really makes that processor work hard.
When you run the program, what you do is type in words and phrases that verbally describe the scene you want painted. Let’s say you type in something like “A white cat sitting on a table with vases of flowers” as an example. What you get is something like this –
MrsGF thinks that image is just creepy but personally I think it’s kind of charming, but everyone’s tastes are different.
But these art AIs have also generated a hell of a lot of controversy as well, and for good reasons.
First of all, artists, actual real world artists who create real world art, are very nervous about all of this and for very good reasons. There are thousands of artists out there who earn their living as graphic artists, illustrators, etc. for advertising agencies, magazines and the like. Publishers and corporations don’t give a fig about things like ‘artistic integrity’ or keeping illustrators gainfully employed. They only care about money. So if it comes down to paying an illustrator money to create a piece of artwork for them, or generating something essentially for free with an AI, guess which option they’re going to take? They’re going to go with the AI everytime.
Another controversy is how these AIs are trained. Before these things will work properly they have to be trained by feeding them literally millions of illustrations. They do this by scraping the internet for every image they can get their hands on, whether those images are protected by copyright or not. Under a strict interpretation of copyright law some claim that this is an illegal use of those copyrighted images. Others claim that this use qualifies as what is known as “fair use” and is legally and ethically acceptable. I’m not a lawyer and I’m not even going to try to figure out whether how these things are trained is legal or ethical or not.
Then there is the question of who “owns” the artwork generated by an AI. That image over there on the right was generated by Draw Things. So who actually owns the rights to that image? You could argue that the image was generated by my phone, using an application that I own, so therefore the image is mine and I own all rights to it.
But it isn’t as simple as that. These days you don’t actually “own” the software you are using. You’re merely licensing it. Basically you’re renting it. That’s been the case for decades now. Under the licensing agreement you agree to, and which you almost certainly haven’t bothered to read because you’d have to hire an attorney to figure it out for you, the maker of the software can set any conditions they like, and it is claimed that your mere use of that software means that you agree to all of the conditions in the license, whether you read them or not. It would be entirely within the realm of possibility for the maker of the AI you’re using to claim any and all rights to any artwork generated by the program. Unless you carefully read the fine print in the license agreement, you have no idea what you are agreeing to.
Some people have started to claim that the AI that generated the image “owns” it. The people who are making this claim either need to find a new hobby or stop taking whatever drugs they’re on because this is just stupid. An AI isn’t an intelligence, it is a computer program. Period. It is no more ‘alive’ than the old Eliza programs people were playing with back in the early days of computing. They’re little more than a complex series of “IF-THEN-ELSE” choices pre-programmed into a computer. Oh, all right it’s a lot more complex than that, but what it boils down to is that an AI is just a computer program like the one I’m using to write this. It is no more “intelligent” or self aware than the calculator you use to figure out your taxes.
The other thing that makes people very nervous is the fact that these things can be used to generate images that mimic specific artists. Want to have your very own original Van Gogh to hang on the wall? You can. These things can easily mimic the artistic style of most well known artists, living or dead. Living artists claim that this reduces the demand for their original artwork. People who like their work don’t need to buy an original or, more likely, an authorized print of the original work, they can use an AI to generate their own in the style of that artist.
I would imagine that the companies that have become wealthy selling stock images for use in publications, companies like Getty, are terrified by this kind of technology because it would very easily put them out of business. Let’s say you’re writing an article about, oh, shopping malls, and you want to put in a photo of a nice looking mall. Instead of going to one of these stock photo companies and buying one, just make your own. Like the one over there on the left. It took Draw Things a little over a minute to generate that. My cost? Zero. So why would anyone ever need to buy generic photos for illustrating articles any more? So let’s just say that investing your pension fund in companies that sell stock images probably isn’t a good idea.
And these things can be used to produce some not so nice images as well. Pornography, s0-called ‘deep fakes’ that portray celebrities, politicians or other well known persons in, oh, compromising positions, let’s call it, can all be generated with nothing but a program like this and a few written descriptions of what you want to see.
Some of the developers of these programs are starting to put filters into the systems to try to prevent or at least limit their capabilities so they can’t be usfed for illicit purposes, but that is ultimately futile. The technology is now out there. A lot of it is open source, meaning anyone can tinker with the underlying code. And there are versions of these programs already out there in the wild, so to speak, tweaked to specifically crank out these types of illicit images. Let’s face it, my friends, that’s what human beings are like. As soon as human beings developed things like writing, painting and drawing, they started using it to make porn.
There are already people out there demanding that something need to be done. Right now. By someone. Somewhere. Congress needs to act now! Act how, though? Ban the software? Sorry, that horse is already out of the barn. Thousands of people are already using it and hundreds are tinkering with the code to make it work even better.
What it all boils down to is that we have a lot of questions, a lot of ethical concerns, a lot of legal issues, and no real answers.
Please feel free to leave comments about this. I’m curious about what other people think of all of this and if anyone has proposed solutions to the problems.
grouchyfarmer 
Incredible tech.
Are we getting to the point where the idea to make the picture, is what can be copyrighted?
I don’t know how I haven’t found your blog yet… But I’m here now. 😉
LikeLike
Thanks for stopping in! 🙂
I don’t think anyone is really sure yet how the copyright situation is going to ultimately be resolved. There is a distinct possibility there will be lots and lots of lawsuits in the near future of the technology.
Right now everyone seems more concerned with the fact that the millions and millions of images that are use to train these AIs are being used without the consent of the artist/photographer who created them in the first place. I think that is a legitimate concern but with the caveat that these days you can be sure that if you post an image anywhere on the internet it will almost certainly get used for something by someone if not outright stolen.
We are just going to have to learn to adapt. The technology is out there right now and nothing state legislatures or congress can do is going to be able to pull it back.
LikeLike
I think that plagarizing the entirety of an art work will always be a copyright in infringement but I also think that parodies are not … and as to lifting elements (bits or pieces) of an artwork and incorporating it into something else via AI would be hard to litigate because bits and pieces are not representative of an entirety of a work and can be found anywhere in Nature.
LikeLike
Oh, wait… amateur astronomer? I have that very same problem.
What kind of gear you got?
LikeLike
My big scope is a Celestron CPC 1100 XLT, an 11″ Schmidt-Cassegrain with the built in computer and GPS system. It’s an absolute gem of a scope. Alas it also weighs something like 70 lbs and dragging it around and getting it set up is really a 2 person job. Between that issue and light pollution here where I live trying to use it got to be more frustrating than anything else. My son has had it now for about 2 years. He and his wife lug it up to the shore of Lake Superior where the sky is really dark.
LikeLike
Nice scope. I have the Celestron 8″ SCT riding on an ASGT mount. That was my first scope.
My big gun is a Discovery 12.5″ solid tube dob. Just killer. Easiest big scope I have to use despite its size. With the aid of a hand truck I can have that scope out the door and setup in less than 5 min, with minimal effort.
Also have an ED 80 on a home built alt/az mount. Very nice Milky Way sweeper. I can push it to around 200x on a good night. Nice little scope.
Also have various binoculars ranging from 20×90 to 10×50. I used to build paralellograms and sell them, so I have a couple of those to play with.
I have bortle 4 skies out my front door, which is awesome. But also have neighbors who think the opposums need their porch lights to see at night. So that sucks. No amount of talking to them helps. Hard headed redneck country.
It didn’t take me long to realize how much work is required to setup the SCT. Being a lazy fucker Iusually use the dob. Its simplicity, in setup and use, is so much faster/easier that if I plan a serious session, I go for the dob.
Of course quick looks, the ED 80 is the go to.
“up to Lake Superior” That sounds like you could be in Illinois…? I grew up near E. Peoria. Still have friends there.
LikeLike
Heh, my dumb ass should have looked at a map. In my shallow defense, it’s been a few years since I tested on the Great Lakes in school, or even looked at a map of the area. Let’s shoot for Wisconsin being S of Lake Superior.
…sigh.
.
LikeLike
Yep, that would be reasonably accurate . Although even Lake Superior would be hard to find right now with all the snow.
LikeLike
The AI image generation technology will, indeed, cause waves throughout the art world because with it, every individual can be their own artist. No more need to spend money on someone else;s labor. From that standpoint, it will not be so good for certain economies.. economies such as the Creative Arts world. I am sure AI will eventually permeate the movie industry and the broadcast media and the publishing industry and b because of that, it will cause a seismic shift in the paradigms to which it is exposed. No doubt about it. It is a massive game changer. It might also, someday become a threat to certain factions of the general social order as well. Anything can be weaponized. I am thinking the porn industry will welcome it. Sooner or later, some AI programs will be able to replicate images of actual human beings and give them such roles as broadcast talking heads, weather people, news anchors and so on. It is a frightening step albeit a fascinating one as well.
LikeLike
You’re spot on with all of that. This is going to be extremely disruptive technology in many different ways. The K-Pop industry is already using the technology to create idols that are entirely computer generated.
LikeLiked by 1 person
Thank you for your most welcome comment, Grouchy Farmer — please visit me often as I will you.
LikeLike
Elon says that artificial intelligence is the most dangerous threat there is to human civilization and that it is advancing too rapidly and that no one is regulating it and that, once weaponized and having fallen into the wrong hands, can be devastating to whoever uses it for nefarious purposes.
LikeLike
I think that this is one case where Musk is right. A lot of people are dismissing the concerns but anyone with half a brain can see that the potential for abuse in the area of AI is going to be an extremely serious issue. AI creation of photorealistic images, the altering of existing images, etc. is out there in the wild, so to speak, right now, and it is getting so good that it is nearly impossible to tell what is real or faked. An acquaintance of mine with some of these tools showed me a photorealistic image she’d made of a prominent obese politician being, oh, let’s call it being overly friendly (cough cough) with a chicken.
LikeLiked by 1 person
You made me pee my pants –LOL
LikeLike
I have been experimenting with AI for writing but I haven’t graduated to art yet, I find the whole concept to be fascinating, It is very helpful.
LikeLike
The problem with AI is that while they can be impressive, ultimately there is nothing really creative or original in the work they produce. I used to work with these systems long ago, even experimented with writing my own conversation type AI. There is no actually intelligence at work. What they’re doing is taking information from a massive database of other people’s work, picking and choosing specific elements from that database according to some set of rules, pasting it all together and then spitting it out. I have a great deal of sympathy for real artists who claim that these things are simply wholesale plagiarism machines stealing real world persons’ material and regurgitating it in an altered form, because that’s pretty much exactly how they work.
LikeLiked by 2 people
As to the written word, such as in blog posts, everything that is written on any blog post has its origins in something that has been expressed at one time or another somewhere in the world and is then transposed into the idea patterns and style of the one writing the blog post. It would be a daunting task indeed to try to write something that has not been written in one form or another at some time or another and I personally think such a task would be impossible.
So I take the AI generated material and refine it to match my own knowledge base and my own biases, perspectives, observations and opinions, and what comes out is pretty much me …at least that is the impression I get …Art, of course, is another discussion and I agree with your perspectives on art and its relati onship to AI one hundred percent.
LikeLike
Considering the quality of some of the blogs I see out there you do have a point about them being unoriginal.
LikeLiked by 1 person
Contradictorily, is there anything in life that is “original”? 😎
LikeLiked by 1 person
Now that’s an interesting question. That one question could make an entire PhD program in philosophy.
LikeLiked by 1 person
As to the problem of artificial intelligence plagiarizing text works, I am given to believe that the AI program converts the written material to another form and even learns to match the way the writer thinks after a while, so in the end, written material becomes more or less original despite where it first appears. That is the first thing. The second thing is that there are plagiarism detectors available on the web that will analyze written material and tell the writer whether it contains plagiarized material or not. Some of these checkers will even show where the plagiarized material comes from.
LikeLike
I’ve just started using ai for book covers, and while the result isn’t that professional, yet, it’s cheap and gives me exactly what I want. It’ll have to get more expensive, I think, and the copyright issues will need to be made clear… either the software owner or the image creator. Stockphoto sites already sell both kinds of copyrighted image.
LikeLike
I think that’s a good idea. AI art can be rather easily designed to suit a specific image that an author wants to convey for a book.
The copyright thing, though… AI is a tool. Period. Claiming that an AI “owns” the copyright to a work of art is the equivalent of trying to claim that the pencil I use to draw a sketch owns the copyright of something I draw. The person who uses an AI to create a work of art is, or should be, the copyright holder, not the tool used to make it.
LikeLike