CARRBORO, N.C. (WNCN) — Remember when most polls had Hillary Clinton winning all the way up to Election Day? She didn’t. How about a similar situation in Britain? The polls said Brexit wouldn’t happen. It did.
“The main thing is if you really want to understand who people are and what they’re thinking and feeling, you don’t ask them. It’s not that everybody’s lying, it’s that even when they’re trying to tell the truth sometime we are just poor witnesses to our own narratives,” said Richard Boyd, CEO of Tanjo Inc.
The Carrboro, N.C. company says it has developed an alternative that bypasses one of the reasons pollsters get it wrong. People don’t tell the truth, even if it’s unwittingly. Boyd said pollsters also use out-of-date tactics like focus groups, surveys and interviews.
“If you really want to understand what people think and feel, you look at what they do with their attention and with their money,” Boyd said.
“So, what did predict [correctly] the last election or Brexit? It’s Google search history, it’s Amazon purchase history, what happens with location services which we all have on our phones. That data exists that we’re all giving off and tells us this sort of life story, this sort of pattern of life which we can then look at to see what is actually likely to happen,” Boyd added.
Boyd said artificial intelligence can predict how communities will benefit from things like the CHIPS Act or the Inflation Reduction Act and even local school bonds.
When others were just looking at COVID-19 infection rates, Tanjo modeled all 3,000 U.S counties using people’s online behavior.
“And [we] looked at things like what are people’s values, traditions, who are their influencers and how are they likely to react to different policies like social distancing and mask wearing and then model that on top of infection rates and that gives you predictor of what is likely to happen,” Boyd said.
Tanjo was pretty spot on, and Boyd feels confident the same will be said when it’s applied to politics.