Elon Musk remains baffled by the process of creating images through artificial intelligence.
Elon Musk's AI Art Debacle
Elon Musk recently finds himself in hot water over his claims regarding X's AI bot Grok. Musk enthusiastically shared a series of AI-generated images, showcasing people amidst musical instruments. Upon closer inspection, these images were not entirely original creations of the AI, but rather, lifted from real photos, all captured by the same renowned photographer, Charles Brooks.
Posting these images with the Grok watermark, Musk told his staggering 220.8m followers to "Generate images with @Grok." However, the quality and inventiveness of the compositions raised suspicion among Grok users, leading to questions like, "Why when I ask Grok to generate images do they look like a blind 3-year-old drew them?"
Upon further examination, it turned out that the images were made using Charles Brooks' photos as references. The Grok user who generated the images uploaded Charles' photos and asked Grok to add the people, the only new element added to the compositions.
Peering into Generative AI
Elon's blunder raises questions about his understanding of generative AI. Chris Brooks, the cellist and photographer, explains in his video response that Grok couldn't have generated those images without using his photos as direct references.
Generative AI for image creation differs significantly from simply copying or referencing specific images. It synthesizes new visuals based on patterns learned from vast datasets, not by directly copying from a reference[1][2][3]. The AI interprets given text prompts and generates matching images by combining visual concepts and features learned during training[4].
While the output is original, it can still be influenced by the styles, compositions, or biases present in its training data[5]. This emphasizes the intricate nature of generative AI, blending originality with learned elements.
The Battle over Originality
Despite generating images based on patterns rather than copies, ownership, copyright, and right-to-use concerns remain complex and evolving in the realm of AI-created art[6]. Inadvertent replication of styles or features closely tied to specific artists or copyrighted works is a risk, especially if the AI is explicitly prompted to generate images "in the style of" a particular artist[4].
Chris Brooks, while accepting that people can edit his images, urges honesty and the use of visible credit[7]. Unfortunately, when he responded on X to point out that the images were not AI-generated, his comments were hidden by the platform.
This isn't the first time Musk has faced trouble with AI imagery. Last year, the makers of Blade Runner 2049 sued Tesla for using apparent AI imagery inspired by the film[8].
As we delve deeper into the world of AI, it is crucial to follow proper protocols and strive for transparency to avoid entanglements arising from misunderstanding the fundamentals of generative AI.
- The incident with Elon Musk's AI bot Grok has sparked a discussion about the understanding of generative AI, as illustrated by Chris Brooks, the cellist and photographer, who questiones Musk's claims.
- Generative AI for image creation differs from direct referencing or copying of specific images, instead synthesizing new visuals based on patterns learned from vast datasets, interpretating text prompts, and combining visual concepts and features.
- While the output of generative AI can be original, it can still be influenced by the styles, compositions, or biases present in its training data, emphasizing the intricate nature of this technology.
- Proper protocols, transparency, and respect for the ownership, copyright, and right-to-use concerns are vital in the realm of AI-created art, as misattribution or inadvertent replication of styles can lead to complex and evolving legal issues.
