Silicon sampling undermines public polling

Some companies are replacing human respondents with AI 'answers' (silicon sampling), and bad actors can use AI to automate fake survey responses at scale; both practices distort estimates and exploit opt‑in survey models while probability‑based panels remain more resistant. The issue is methodological (how surveys are conducted) but also civic: if poll results reflect algorithmic stereotypes or fraud, media, policymakers and markets will be misled. — If AI substitutes for or systematically corrupts public-opinion data, the feedback loop between citizens and institutions — policy, elections, media narratives — will be degraded.