In an op-ed for Le Monde, a group of economists and creative-industry experts examines the emerging tensions between generative AI operators and authors, recalling that “artificial intelligence works like an oil well—and that oil is our creations.”
A video of a spectacular fight between Brad Pitt and Tom Cruise has been circulating online for the past few days. Yet the scene was never filmed. Neither actor was present. And yet their faces, voices and movements are all there. It was generated by a generative artificial intelligence (AI) system from a two-line prompt [request].
Hollywood is sounding the alarm: will its actors and directors one day be replaced by their own avatars? Those who write screenplays, record their voices or compose music have already seen their work, for the past two years, being absorbed by machines capable of imitating them. Coachmen disappeared with the advent of the automobile. But they had twenty years to become drivers. The professions threatened by AI do not have that kind of time. The production of the expertise on which they depend could dry up within a matter of months.
Artificial intelligence works like an oil well. Except that the oil is our creations: films, images, books, translations, software, articles, comments and expertise. This is the reservoir from which the models draw. If no one is paid to keep replenishing it, we will soon be left with nothing but the same frozen stock, endlessly recycled.
Imposing collective management
This problem is not new. At the beginning of the 20th century, the Texas oil wars triggered a race to extract resources that prematurely exhausted fields that were, in fact, still abundant. Pumping too quickly caused pressure to fall. The oil stopped flowing—even though it was still there. Collective management had to be imposed in order to maintain the reservoirs and control the pace of extraction.
AI faces the same challenge: what will become of these systems if they are forced to rely forever on the same data? This is especially true of the most sought-after data, since not every drop is equally valuable.
To generate a video featuring Brad Pitt and Tom Cruise, anonymous faces are not the most useful source material; what matters are precisely the faces of those two actors. Their manner of speaking, their features and their gestures are far more valuable fuel for achieving the desired result. Likewise, a photograph of the Mona Lisa is no more valuable than an ordinary portrait—unless the desired output must be produced “in the style of Leonardo da Vinci.” In that case, every brushstroke analyzed in the original work becomes an invaluable resource.
Conversely, AI models would be ineffective without the abundance of more commonplace content. As in an oil field, every drop matters. But unlike an oil field, every drop has its own value. This is where a new industry is beginning to take shape: the industry of valuing intellectual oil. The technology is ready, but meters, standards and markets still need to be established so that differentiated value can be assigned to material that, even today, is still being captured wholesale.
A collective management organization
Europe may possess the world’s largest reservoir: 450 million educated citizens producing texts, images, code, knowledge and culture every day. Yet it owns almost none of the wells. Its interest, therefore, is to develop, in cooperation with AI operators, an industry capable of measuring the usefulness of each contribution and remunerating each contributor accordingly.
As AI is expected to become the leading driver of the economy, we may be entering a transformation of historic proportions, built on intangible assets and intellectual property. It would be absurd to allow innovation to cause a global slowdown when the opportunities are so vast.
To ensure that this valuable fuel is fairly valued and that authors, creators, performers, producers and publishers are properly remunerated, we propose creating a Europe-wide collective management organization, based on a model inspired by the French Society of Authors, Composers and Music Publishers (SACEM).
In this way, we will give rise to this new industry, for the benefit of Europeans—but not only Europeans. All the major beneficiaries of generative AI, and especially the giants that feed on this fuel, will pay a fair price… at the pump.
Jean-Paul Betbèze, emeritus member of the Cercle des économistes
Maurice Lévy, founder of VivaTech and Chairman Emeritus of Publicis Groupe
Vincent Lorphelin, founder of the Controv3rse think tank
Christian Saint-Etienne, Professor Emeritus of Economics