{"version":1,"type":"rich","provider_name":"Libsyn","provider_url":"https:\/\/www.libsyn.com","height":90,"width":600,"title":"8-3-26","description":" Las empresas de inteligencia artificial est\u00e1n comprando millones de libros usados, escane\u00e1ndolos para entrenar sus modelos de lenguaje y desechando las copias f\u00edsicas. A medida que el contenido de Internet se genera cada vez m\u00e1s mediante inteligencia artificial, resulta dif\u00edcil actualizar modelos ling\u00fc\u00edsticos como ChatGPT, ya que la informaci\u00f3n con la que se alimentan no es de origen humano y, por tanto, es de baja calidad. Empresas como Anthropic se apresuran a adquirir la mayor cantidad posible de libros, especialmente los m\u00e1s antiguos, para garantizar que el texto no haya sido generado por IA, con el fin de entrenar sus modelos. Sin embargo, para escanearlos, primero deben destruir la encuadernaci\u00f3n e introducir las p\u00e1ginas en una m\u00e1quina. Esto implica no solo la p\u00e9rdida definitiva de libros raros, valiosos y de gran importancia cultural, sino tambi\u00e9n que su contenido ser\u00e1 propiedad privada de las empresas, las cuales pueden retener la informaci\u00f3n o modificarla antes de publicarla. Recientemente, Anthropic pag\u00f3 mil quinientos millones de d\u00f3lares para resolver una demanda por el uso de libros digitales pirateados en el entrenamiento de su IA. Este proceso contrasta con los esfuerzos de las bibliotecas por digitalizar libros para facilitar el acceso p\u00fablico a ellos. AI companies are buying millions of used books, scanning them to train the artificial intelligence language models, and then throwing away the physical copies. As content on Internet becomes more and more AI-generated, it becomes difficult to update language-based AI models, such as ChatGPT, because the knowledge being fed to it is not human-made and is therefore low quality. Companies like Anthropic are rushing to buy as many books as possible, especially older books to ensure that AI was not used to generate the writing, to train their models. However, to scan the books, the companies first have to destroy the binding to feed the pages into a machine. This means that not only are rare, valuable, and culturally important books being lost forever, but the writing in them will now only exist under the private ownership of corporations, who can withhold the information or change it before publishing it. Recently, Anthropic paid a one point five billion dollar settlement for using pirated digital books to train its AI. This process is in contrast to efforts by libraries to digitize books for public access. ","author_name":"Bolet\u00edn de Noticias de KBBF \/ KBBF News Bulletin","author_url":"https:\/\/sites.libsyn.com\/460356","html":"