What emerges from the presentations that the 2023 edition of Forum Edita has dedicated to the tsunami that is taking place in the creation of content by artificial intelligence (badly called that because they are neither intelligent nor artificial) is that progress has caught us with the wrong foot and perhaps little prepared for its unstoppable advance. The conferences that are celebrated in the Pompeu i Fabra University of Barcelona Until this Wednesday, it has brought together different specialists in the field from different perspectives, but mostly around how this new, or not so new, technology (because it has only expanded its capacity and power to combine previous data) can transform the current model of publishing and its forms of work, in addition to what it means for creation and the sacrosanct copyright.
The writer opened fire Jorge Carrionwho despite having several analytical papers on the subject and having used the GPT-chat As an instrument, it has decided to withdraw in the future into a more artisan and analog type of creation, which does not invalidate, in its opinion, the potential that AI has as a creative tool. An example, a novelist can preselect the characteristics of a personheh and ask him to create a whole ‘background’ for himself that will help him when writing. “This is not very different from what Juan Goytisolo did when he was barely a teenager, he cut out photographs from magazines to create the characters for his stories & rdquor ;. Carrión is more reticent about the imaginative capacity of the AI -imagination is not something that can be assumed for a mechanism that is governed from training and a combination of previous data- and perhaps now is not the time but it is possible that in the not too distant future, according to Carrión and without giving names, machines are capable of “writing better books than some best-selling writers” and retire “90% of the authors who work with marked structures & rdquor ;.
Two scripts to choose from
How will AI influence the publishing industry? In many and varied ways, but it seems highly probable that work like translation, especially the technical translation, not literary translation, come accompanied by the help of the machine. “Today, you don’t pay the same price for a translation in which AI is used as a base than for an ‘original’ one,” says Carrión. For him, however, the creation of content by AI and traditional creation will have a long coexistence because “the paper book and the teacher & rdquor; they are unbeatable”.
The most technical approach to the subject was provided by the computer engineer Josep Maria Ganyet, seconded by Javier Aparicior, one of the directors of the Forum. Ganyet reminded those who get carried away by apocalyptic ghosts the famous phrase of the late Alan Turing: “Having intelligence is not the same as having intelligent behavior”, which would indicate that machines are competent but not intelligent and of course their ‘knowledge ‘ Let’s put it that way, it has little to do with magic. “They can be creative because they know how to copy, combine information and transform it, but they lack something very human, which is the objectives.” How Pablo Picasso summarized in a pre-technological time: “computers don’t know how to ask questions& rdquor; Or what is the same, artificial intelligence is an engineering machine and not a thinking being.
Need for ethics and legislation
Faced with fearful and doomsayers, Ganyet postulates an ethic that controls the entire value chain of the creation of ‘artificial’ content that must be validated by people, an ethic that had its concretion in the ‘declaration of Barcelona’ of 2017, created within the framework of the International Center for Scientific Debate in which prudence, reliability and responsibility of companies and institutions were requested.
Related news
Another major unresolved problem to date is the copyright issue. The regulations used up to now could not foresee the great potential of the current content generative models that are not a mere cut and paste but use ‘cured’ data, that is, selected and validated so that they can be used by the algorithms. The big problem is that there is currently an open bar with respect to these original data and there is the paradox that computer content is competing with humans, which is why legislation, not yet specified, is needed to obtain copyrights regarding those raw data. The contributions of MEP Ibán García del Blanco, one of the European politicians who is working the most on the subject, circulated in this direction of intellectual property, who notes that Europe is currently leading the defense of these rights against a more lax United States. It is only necessary to appreciate that one of the demands of the Hollywood actors’ strike refers to legislating the use of their images by AI.
Jorge Corrales, general director of the Spanish Center for Reprographic Rights (CEDRO), insisted on the need for negotiation and authorization by the AI of the original data, as well as the differentiation of content created by machines or by humans. He also warned of the bias and impoverishment of the fact that only English is the language used to train AI systems.