“AI generates images 1”
Image extraction: DALL・E2

Addendum (2024.06.12):
It has become necessary to change the method of image generation from the ones listed below. For more details, please refer to this page.

—

If it were an investment-related blog, it would be usual to use images that correspond to the content of the blog, such as photos of banknotes, pictures of working bankers, or the exterior of stock exchanges.

However, in my personal opinion, in many blogs, images are only considered as mere additions to the text, like cherry tomatoes in a lunchbox, and aren’t given much importance. I believe that bland images made primarily for SEO purposes, which are forgotten about just five seconds after being seen, are proliferating.

So, in this blog, I decided to focus heavily on my own hobby.

I have an interest in art, mainly painting, and the timing when I started this blog was at the dawn of generative AI, so along with testing the accuracy of market predictions using ChatGPT’s sentiment, I’ve been trying out the generative power (if that’s what it’s called?) of Dall E2, Stable Diffusion, Midjourney, and others. I’ve been posting images with the concept of an "art gallery."

Generative AI Art Gallery

As a person with personal tastes inclined towards Matisse, Picasso, Modigliani, Paul Klee, and even Warhol, Basquiat, Francis Bacon, David Hockney, and especially in a league of his own, Willem de Kooning, I have been selecting artists and generating AI images for exhibition using my ample prejudice and discretion. The images on this site are based on the concept of subjects related to finance and investment.

Renoir, Manet, Monet, and other Impressionists like Dali, Jackson Pollock, and Keith Haring might be considered more popular artists in Japan (?), but I’ve dismissed them as I lean heavily into my personal interests, as previously mentioned.

Van Gogh, Klimt, Banksy, Marlene Dumas, Tsuguharu Foujita, Taro Okamoto, and Ukiyo-e artist Kuniyoshi Utagawa are among the artists I would like to try out in the future.

I also like Mark Rothko, but it’s clear that the generated images will eventually have the same usual feel, even when the concept is finance, so he is excluded. The same goes for Turner.

Ignoring eras and styles completely, I am generating AI-created images in the style of artists I want to see at that moment, almost randomly, and if they look good, I post them.

I’d like to choose a wider range of artists, not just sticking to the super major ones as mentioned above, but there’s a problem that the AI hasn’t learned (the generation is strange) if I do so.

Until a few months ago, even Bacon and Hockney didn’t seem to have been learned by Dall E2 and Stable Diffusion. Even if such major artists are like this, it seems quite difficult to use the taste of artists who have not gained fame beyond national borders.

The Problem with Generative AI

Incidentally, this could be said to be a problem, or rather a challenge, that generative AI must overcome. I think it could inadvertently foster uniformity, which is the opposite of diversity.

The information that AI deals with can be selected based on certain measures such as recognition, and other elements may not be uniformly considered, even if those elements strongly relate to or affect specific things.

For me, Francis Bacon and David Hockney are stunning artists, and I cannot imagine a life without seeing their works. However, if AI does not incorporate their information for a certain period of time, they could be forgotten, and the opportunity for their works to reach people with similar sensibilities could even disappear.

To put it simply, I think that what is spoken about by AI can become “good”, and what is not picked up becomes “bad”, and this could become the norm in society. This is predicated on AI acquiring such a significant social impact, but it seems certain that we are at least moving in that direction.

Regarding Generating Images in the Style of Someone

There is, I believe, an ongoing lively debate (as of June 2023) about whether it is a problem for AI to generate images in the style of someone, whether it constitutes plagiarism. I think there are points I can agree with in both the pro-side that approves of it and the anti-side that argues it should be regulated.

That said, given that I actively post AI-generated images in the style of certain people on my blog, I think I would be classified as a supporter, and indeed I am.

However, to add a little more to that, regardless of the pro/anti position, I feel this trend is a game-changing event that cannot be stopped. The position of so-called past works, including art, has shifted from its previous position, and I feel changes are beginning to occur in the very meaning of their existence.

Nevertheless, that does not mean it is okay to neglect current issues such as the presence or absence of copyright. What I can do is simply to use image-generating AI within the limits allowed, dutifully following whatever is decided by the appropriate agencies or nations.

Well, if I am to be honest, putting aside such complicated issues, I simply want to witness something I’ve never seen before, a work in any form – that’s the truth of the matter.