Neural networks are trained on millions of works without the permission of their authors
Over the past year, neural networks that create images, such as Stable Diffusion or DALL-E 2, have received wide coverage in the media and on thematic resources. Despite the fact that the capabilities of neural networks are still limited and cannot be compared with the capabilities of a professional illustrator, the software can generate pictures in huge quantities, and, as you know, quantity turns into quality over time: among the creations of neural networks, there are also real masterpieces.
Some art communities and platforms for artists and designers have decided to ban the placement of images created by neural networks on their resources: DeviantArt and ArtStation have not yet made such statements, but, for example, Fur Affinity has banned the publication of images created by software “in the interests of the community.” Some artists also expressed their dissatisfaction – it turned out that neural networks were trained, including on their works, without the consent or notification of the authors. Experts point out that the mass generation of images by neural networks can lead to blurring the very concept of graphic art and underestimating the work of artists, designers and illustrators.
Write to our site and earn
20 000r discount for 3070 Ti = for a penny
3070 Gigabyte for 55 tr in Regard
3080 Gigabyte for 70 with start
3060 Ti for 40 tr in Regard
3060 is cheaper than 40tr in Citylink
Computers cheaper than 10 tr in Citylink
However, there is an alternative opinion, according to which progress and automation should not be stopped by moral or pseudo-moral attitudes – according to supporters of this approach, if this or that image is available for viewing, it is quite ethical to train neural networks on it, despite the presence of one or another license or copyright. On the web, you can see a wide range of opinions on this issue, and, as experts in digital law predict, in the coming years we will certainly see lawsuits from artists against training neural networks at their work – it remains only to feel sorry for the judges and experts who will have to understand the nuances of legislation which does not always keep pace with technology.
This material is written by a visitor to the site, and a fee has been accrued for it.