Programming

Compensation for those who provide data to train AI: what does OpenAI think?

Compensation for those who provide data to train AI: what does OpenAI think?

One of the biggest doubts related toArtificial intelligenceby its very nature, are the training methods.

In fact, for these models to work correctly, they must be powered with a huge amount of datatextual and otherwise.

These are often covered by copyright and, although the large companies in the sector have often overlooked this issue, they are now increasingly faced with the consequences of their actions. Beyond legal issues, it is legitimate that aGenerative AI come ChatGPT Do you reward those who contribute to your training?

The vice president of consumer products at OpenAIthat is to say Peter Deng, he managed to get around the answer with a certain elegance. On the occasion of SXSW Conference & FestivalsDeng stated how this is “An excellent question” without going into further detail. If the situation, in this sense, appears delicate for ChatGPT, it is even more so for FROM-E 3.

Data to train AI and copyrights: a delicate situation to say the least

Models like DALL-E 3, in fact, are fed with an immense amount of examples, ranging from simple ones illustrations until to works of art. Not only that: the AI ​​also draws on photos and images from public sites and from data sets, the provenance of which is not always so clear as regards the Copyright.

OpenAI, as well as other companies in the sector, support the philosophy of fair use, which allows the use of copyrighted works to create what is considered a “secondary work” as long as it is transformative, without owing any compensation or credit to the original artist. On the other hand, it is true that it would be impossible to create AI models without drawing on copyrighted material.

On the other side of the fence, those who hold the copyright think in a radically opposite way. In this sense, the cases brought to court are numerous and involve an ever-increasing number of artists aligned against tools such as the aforementioned DALL-E 3, Midjourney and similar.

Compared to this thorny situation, Deng and OpenAI limit themselves to a shrug of the shoulders, without delving into the topic much further. For the OpenAI employee, in fact, artists should have more freedom of action in the creation and use of generative AI tools like DALL-E, but he is not sure exactly which formulas can satisfy both parties in cause.

I believe that if we can find a way to make the flywheel of artistic creation faster, we will really support the industry… In a way, every artist has been inspired by artists who came before them” Deng finally concluded.

Leave a Reply

Your email address will not be published. Required fields are marked *