”The problem with generative AI has always been that … it’s statistics without comprehension.”
—Gary Marcus
https://garymarcus.substack.com/p/new-ways-to-corrupt-llms
#ai #generativeai #llm #llms
Follow
@janhoglund a few years ago, researchers at Google documented a similar list of leakages in their then-new image generator model enriched with linguistic semantics (see Section 6.3). I like it a lot, because it demonstrates similar issues visually.
https://arxiv.org/abs/2206.10789
