The source is a blog post by Enrique Dans titled "Cuando juzgar a la inteligencia artificial generativa por lo que sabe hacer hoy es un completo error," arguing that evaluating generative AI solely on its current capabilities is short-sighted, comparing it to judging early aviation by the Wright brothers' initial flights.

Dans discusses the limitations and progress of generative AI in complex tasks like programming and academic research, referencing critiques from developers like Thomas Ptacek and his own experiences using AI for structuring academic papers versus the actual research.

The post also highlights the evolving role of AI in academic publishing, from generating initial drafts to assisting with peer review, suggesting a future of "academic centaurs" who leverage AI to enhance human capabilities rather than be replaced by it.

Ultimately, the article emphasizes that AI is rapidly advancing, and its future potential extends beyond its current linguistic focus, requiring a conceptual shift to understand complex systems.

You can also read this article in English on my Medium page, «Beyond the finish line: generative AI’s unknown potential«