AI and neural networks (Midjourney, DALL-E) paint beautiful aesthetic pictures on request from a few keywords. Is the author AI or a human who wrote these 2-3 words?
Since it is not a very easy and straightforward answer to this question I can answer with an example: When a photographer takes a photograph who is the author of this photo, the camera or the photographer? I say both. So if we assume that a camera is a tool and human is the user the credit is given to the user. Similarly if we assume that the AI image generator is the tool and a human is the user then the user can be credited as the author. But this comparison is not true because a photographer needs training and learning skills to take a good photo and he/she probably can repeat that good result many times. However in AI image generation 99.9% of the job is done by the algorithm and the user has a minimal control over the results.
this is indeed a difficult question that lies in the realm of the philosophy of art and ethics. And these questions need to be raised now. Maybe such paintings deserve the definition of "collective art", that is, one that is human property and does not belong to anyone
Yana Suchikova
Following the analogy of the photographer, which I think is easier get a concise analysis, the camera is a tool, not the creator of the work. The most immediate reaction to uphold the camera as creator is to state that there are auto focus algorithms in the camera that uses "intelligent" algorithms to take a better quality picture. Yet, this contribution is not creative and this makes all the difference.
The way that creativity can be described in my opinion is best exemplified by the 'non obviousness' concept used in patents. Now, unless the camera comes up with a 'non obvious' way to do auto focus, then I would not consider it a creative contribution to the work, just a tool. The immediate counter argument would be that the artist is not being creative either. The answer would be that, yes, photographers can make instances that are not creative and just mechanical. This is what distinguishes a good work of photography as art (besides aesthetics).
The output of the neural networks are not 'non obvious since they just mimic output that stays within a predefined distribution. Jumping outside this distribution in a coherent and substantial way is what makes a creative work and what a machine learning(ML) algorithm cannot do (by definition of ML learning theory).
Yana Suchikova The authorship of the artworks made by AI is open to interpretation. Some claim that AI is the author since it makes creative judgments and produces unique results. Others say that the author should be regarded as the person who created the keywords used to make the image, or the designer of the AI itself. Finally, the answer may be determined by the context and interpretation of the idea of authorship.
- Badges
- Science topic
- Similar topics
- Methodology
- Laboratory