Cooper Sharp Proves That American Cheese Can Be Great Cheese
John Moltz, in his weekly members-only column for Six Colors (no gift links, but Moltz’s column is worth the subscription in and of itself): This week Anthropic announced that in order to identify AI from human-generated content it would be watermarking all text it generates by subtly altering the randomness at which its models picked the next word. While this might seem a perfectly cromulent…
In a recent development, Anthropic, a leading AI company, announced their plan to watermark all text it generates. This move aims to distinguish AI-created content from human-generated text. However, this strategy has sparked controversy among tech enthusiasts, particularly John Gruber, who finds the approach misguided. Gruber likens the idea to adding autocomplete Swiss to a Philly cheesesteak, suggesting it would only make the product worse.
He believes the EU regulation driving this change is ill-considered and outdated. Instead, Gruber argues that the real motivation behind this decision could be more about protecting their own models from training on AI-generated content, to prevent "model collapse" and "AI inbreeding". Moltz, another tech commentator, echoes these sentiments, suggesting that Anthropic and Google are more interested in watermarking their own text to maintain control over future model training.
Written by urgent.news from Daring Fireball's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
