Previously, in the second half of 2022, with the emergence of a large number of high-performance AI painting tools such as Midjourney, Stable Diffusion, and NovelAI, everyone saw the utility of artificial intelligence in the consumer field for the first time. Since then, the impact of AI on painting and photography has quickly become the focus of discussion among painters and videographers around the world.
After waves of discussion, the industry has gradually formed a consensus that the role of AI in these areas may only be "assisted" rather than "replaced", and that AI is still only a driven tool. However, a recent data released by the overseas research team Stock Performer has come to the conclusion that almost every artist and photographer is terrified, that is, AI-generated images actually make more money on Adobe Stock than human-created images.
Adobe Stock is a royalty-free image library owned by Adobe that provides millions of high-quality, royalty-free, professional photos, videos, illustrations, and vector graphics for designers and businesses to build their creative projects. As the world's second largest image library, Adobe Stock is also one of the few galleries to officially accept AI-generated images in the face of controversy over AI copyright.
To this end, Adobe Stock also sets a relatively strict rule, that is, before submitting relevant content, users must abide by the terms of use of AI tools, own the commercial copyright of the corresponding generated content, and cannot submit works depicting real locations, identifying third-party assets (such as LOGO) or real people (unless authorized), and must clearly use the words "generative AI" or "AI" in the title to identify.
According to data released by Stock Performer, after analyzing Adobe Stock data from 2022 to the present, they found that the revenue per download (RPD) of AI-generated works is 1.59 times that of non-AI sources, and the average monthly revenue (RPI/m) of AI-generated images is 4.5 times that of non-AI sources, with the former being $0.17 and the latter being $0.0375.
In fact, this number is already very telling, after all, this is the result of users taking real money, and it also proves that the content generated by AI at this stage can already meet the needs of some users.
Therefore, it can only be said that the outside world may underestimate the speed of AI technology progress. Previously, last fall, Midjourney and Stable Diffusion needed to be tweaked repeatedly to get a working image. The pain point of AI painting at that time was that the keywords used could not generate the picture that the user wanted, but with the emergence of an AI painting plug-in called "ControlNet" not long ago, this pain point was perfectly solved.
It is reported that ControlNet's painting mode is to first let users enter a reference picture, and then preprocess a new picture according to a certain pattern according to this picture, and then AI will draw the finished product according to these two pictures.
Six months ago, in the portrait drawn by AI, the hand was still an insurmountable difficulty, but it only took half a year to solve this problem. In fact, AIGC replaces painters and photographers, and the progress of AI technology is only a secondary factor, the real reason is "bad money drives out good money", that is, the relatively lower level of AI creation content squeezes the living space of high-level human creators.
So why can painters and photographers become a profession? After all, painting and photography are both skills with thresholds.
Since the user has needs, but does not have the skills to realize these needs, it also constitutes the source of the artist's bargaining power. But the problem is that now with Stable Diffusion, NovelAI, DALL-E, ControlNet, and various Loras, users find that they only need to tell the AI the demand, and it can give the corresponding image. Admittedly, AIGC's level is limited, but a few more attempts can always piece together a barely usable piece.
Although the quality of these works is average in the eyes of real painters and photographers, their bargaining power is weakened by AI, and AIGC's dual advantage of "customization + low cost" is too powerful. There is a saying that Party B hates Party A the most for not understanding anything and blindly commanding, but Party B's bargaining power actually comes from Party A not understanding anything. So the current situation is that AI has become a "spare tire" for some party A.
And one of the most critical factors in an industry from sunrise to sunset is that there is no money to earn. That's right, AI "kills" an industry not refers to the moment when AI reaches the level of high-end practitioners, but when the cost performance surpasses low-end practitioners. Because after low-end practitioners are squeezed out by AI, middle-end practitioners can only get the price of the previous low-end level, after all, the cost of AI is too low in comparison.
The impact of AI on the painting and photography industries is not a momentary thing, but a slow process, just as short video dimensionality reduction hits graphic content.