Should we be afraid of generative AI or not? How to use it ethically? Are the creations generated by this process protectable or not? What about the sources exploited in this process? Invited to the round tables organized on this theme by the Fédération de la haute couture et de la mode (FHCM) during the Hyères Festival on October 14, the speakers at the 22nd International Fashion Meetings attempted to provide answers to these questions. Concerns, which increasingly affect brands and houses, while generative AI has been developing exponentially for almost a year in an almost non-existent legal framework.
As generative AI races ahead, there are few or no safeguards – ph Gertrūda Valasevičiūtė on Unsplash
The fascinating debate was not necessarily reassuring because the path to creating a framework around this new disruptive technological tool turns out to be complex. Brussels is in the process of deciding on regulations in this area. Adopted at first reading by the European Parliament in June 2023, the AI Act promises to be the first law of its kind in the world. Proposed by the European Commission, it is currently in the final phase of discussion between the Council and the European Parliament in order to determine a common position. “This is a harmonization, it is the set of rules which must be harmonized at the European level in order to ensure both that artificial intelligence technologies are safe, that they respond to European values and that they can also promote innovation, explains Eric Peters, deputy head of unit in charge of the 2030 Digital Decade Strategy at the European Commission.
“The heart of the proposal is firstly to ban a certain number of purposes for these technologies which are not appropriate, for example the social credit system set up by China, and secondly to point out a certain number technologies or purposes which are high risk and for which there will be obligations which will ensure that the risks are well controlled”, he summarizes. The idea is to regulate the use of generative AI more strictly with transparency obligations and limits for technological models considered at risk.
“The law will probably be adopted in 2025 and applied in 2026, whereas we must tackle it now. How will it be applied, how will the controls be put in place? These are still open subjects that we must think about very precisely. We can quickly let ourselves be overwhelmed by the creative side, but what is important is to face the question in its reality”, observes Laurence Devillers, professor of artificial intelligence at Sorbonne University/CNRS. And to report this anecdote, which says a lot about current issues. While she was invited to participate in a committee on the ethics of AI and expected to meet mainly scientists, she found herself “faced with all the major big tech players from China and the States -United, who wanted to slow down standards.”
The researcher warns against the excitement that has arisen in recent months around these new technologies: “We must denounce the myth of generative AI. In the majority of surveys, people see the machine as an authority We must stop and reverse this belief, be critical. Big tech wants to scare us so that we do not take any action, but it is just a machine, which has neither the notion of chronology nor of space. AI is not capable of making sense. Certainly, they are great tools, but you have to know how to use them.” And to wonder about the future and the impact of AI: “With the generation of synthetic data, there are major consequences. When tomorrow we arrive at 80% artificial data, what will we do original knowledge, which we already have?
Less than a year ago, almost no one had heard of generative AI. Coming into force at the end of 2022, the European regulation on digital services, the Digital Services Act (DSA) intended to regulate the digital world and in particular the web giants, does not even mention the word AI. But technology is exploding. “The challenge we face is absolutely extraordinary, all the players must align in extremely short periods of time while generative AI has only recently appeared to the general public. We are facing a situation where we must be on all fronts, while things continue to evolve. The important thing is to have a direction and to really ensure that we give the right signals to the actors,” recalls Eric Peters.
From left to right, Vincent Fauchoux, Laurence Devillers, Eric Peters – ph D. Muret
In the meantime, generative AI has already arrived on the market without worrying about the rules, developing for example without respecting copyright, as noted by Vincent Fauchoux, associate lawyer and managing partner of the DDG firm, which also advises the operators of generative AI and their users. “By scraping, storing and indexing all kinds of documents, images and videos retrieved from the Web without authorization, these technologies are built on the heist of the century. They are in the original sin from the start,” says -he.
A market worth more than 1,000 billion dollars
“Finding the source in a creation generated by generative AI is an almost impossible operation, because we are faced with an aggregated whole. A bit like a tomato impossible to find in its original form in a ratatouille. We cannot does not detect the original creation which served as its basis. Original sin leaves no trace”, he continues. However, as the lawyer further underlines, “while the systems have not been constituted in their own way, the work created by generative AI is itself protectable, because we do not find the ingredient original.”
How to remedy this paradox? “A first option consists of establishing extraordinary agreements with the actors involved. Another is to advocate traceability at all levels in all production contracts,” says Vincent Fauchoux. The French Society of Authors, Composers and Music Publishers (Sacem), for example, is considering exercising a right of opposition on the use of its authors’ data and instituting an AI tax to authorize machines to use their works of art. for training. Others propose that a royalty be paid to authors authorizing the use of their artistic content by generative AI publishers.
According to estimates cited by Bloomberg, the generative AI market is expected to reach $67 billion in 2023 and $1.3 trillion by 2032, while investments in the sector are also expected to grow strongly. “There is a lot to gain by using these technologies. But we must do it carefully, highlight good practices. We must help each other, pool information to involve humans who are creative, responsible and aware of the side effects”, recommends Laurence Devillers.
As Europe prepares to rule on the question of regulation around generative AI, Eric Peters wants to be reassuring: “There will of course be a whole system of governance with controls, with standards, which that the actors will be aligned behind European values”. Beyond this common legal front, it is also essential to have stronger players in Europe, capable of creating leadership in the field, the speakers conclude.
All rights of reproduction and representation reserved.
© 2023 FashionNetwork.com
#Fashion #luxury #generative #facing #ethical #legal #challenges
