(Ben Sellers, Headline USA) The recent failures of polls to accurately gauge the support for President Donald Trump have cast light on systemic surveying vulnerabilities such as sampling bias, phony poll weighting and other methodological flaws.
But an alarming guest essay that ran Monday in the New York Times called attention to a new trend in surveying that could make human polling failures pale by comparison.
You gotta be shitting me pic.twitter.com/P4oxtu97qA
— Peter J. Hasson (@peterjhasson) April 6, 2026
Data scientists Leif Weatherby and Benjamin Recht warned of “silicon sampling,” a process in which artificial intelligence uses predictive simulations to anticipate what people would say if polled, without actually polling them.
“Because large language models can generate responses that emulate human answers, polling companies see an opportunity to use A.I. agents to simulate survey responses at a small fraction of the cost and time required for traditional polling,” the researchers wrote.
They noted that traditional polling methods had become too difficult or unreliable due to the past failures in the field that have led respondents to approach pollsters with skepticism and apprehension.
The two conceded that problems of bias were rampant in existing polling methods, saying that “pollsters can use modeling to nudge polls in a certain direction and influence public opinion itself, rather than merely to report what the public thinks.”
Nonetheless, the use of algorithms to supplant actual human responses was sure to have one predictable outcome: exacerbating the issue itself.
“The computational whiz kids behind silicon sampling are so excited about A.I. that they will insist that their complex predictive computer simulations are accurate because they are trained on what’s been observed in the past — therefore, they excel at simulating human behavior in the present and predicting what’s next,” the researchers wrote. “However, prediction is not the point of polling. The point is gathering current opinion.”
The article also vaguely alluded to one of the biggest concerns surrounding artificial intelligence: That far-left Silicon Valley programmers may program their own radical left-wing biases — whether intentionally or not — into the systems themselves.
“The further from people we get, the more the simulation becomes a mirror of the pollster’s beliefs,” the essay said, citing a study from Cornell University that silicon sampling made the data even more skewed.
Platforms such as Google Gemini have been slammed for imposing bizarre DEI filters that result in historical revisionism, turning notable figures who were demonstrably and objectively white into minorities.
ChatGPT, one of the leading AI programs, has even openly admitted to programming left-wing bias into its algorithms.
“Since our launch of ChatGPT, users have shared outputs that they consider politically biased, offensive, or otherwise objectionable,” the company said in a 2023 statement. “In many cases, we think that the concerns raised have been valid and have uncovered real limitations of our systems which we want to address.”
Just as some pollsters may be less biased than others, not all AI systems are equally skewed. Billionaire free-speech proponent Elon Musk, who has used X to develop his own Grok AI system, has signaled his commitment to countering bias in the emerging AI field.
President Donald Trump himself also issued a directive to develop bias-free AI shortly after taking office in January 2025.
But in an industry already riddled with complaints that pollsters face no accountability for their failed prognostications, adding another layer of obfuscation to the complex methodologies while cutting corners in the process could ultimately signal the end of polling as we know it.
“If we do not slam the brakes on silicon sampling, we could see a significant undermining of trust in public opinion work and social science research more broadly,” said the Times essay.
Ben Sellers is a freelance writer and former editor of Headline USA. Follow him at x.com/realbensellers.
