Source: Marcus on AI
As AI systems proliferate, the internet fills with low-quality machine-generated content—what some call "slop"—that degrades signal-to-noise ratios. This creates a cycle where AI trains on increasingly contaminated data, producing worse models, while eroding economic incentives for human creators. The productivity gains promised by AI go toward quantity over quality, leaving consumers with fewer genuinely useful resources and platforms unable to separate signal from noise.