The100: Polling, chatbot debacles and shopping wonky

This is what you will believe today 

Monmouth Poll posted some clickbait research showing that 1 in 5 Americans believe that Taylor Swift is part of a covert government effort to re-elect Biden. 

But (and it’s a pretty big one) 42% of those who reported believing this, hadn’t heard about that conspiracy prior to being surveyed. Erm?

As Ariel Edwards-Levy says:

[…]willingness to entertain an idea when asked about it doesn’t necessarily indicate that someone is firmly bought into the idea outside that context.  Both openness to conspiracies and actual IRL [in real life] adherence to them are relevant, but they’re different metrics.

Delegated AI-lments

Continuing Swift-ly on… John Willshire on the Air Canada chatbot debacle. Their chatbot changed a refund policy and the airline denied responsibility. Jon asks an interesting question: How will companies react if every claim their chat bots make is now legally binding?

On saying one thing, and doing the other

Another interesting survey from YouGov, looking at what the US national media focus on and whether that is too much or too little. It says more about America, serious policy debate, and how badly (by inference) we’re all served. Partisan divergence and its absence are also fascinating.

Segmentation segment

Get this: DINKS in the US are doing pretty well for themselves, while Britain has killed off its musical tribes in our age of the ‘big-flat-now’.  

The diminishing return of ‘average’

John Heggarty (the H of agency BBH) has a newsletter and it contains this excellent quote from Tom Goodwin:

“The mechanics of large language models (LLMs) produces likely combinations of things. So you’re getting the average of the internet,” he says. “It’s groupthink, truisms, banality. It’s not wrong, but it’s worthless. The average poem doesn’t make people famous. The average tennis player doesn’t make any money. The average songwriter is a failure. Averageness doesn’t even deliver average results. We are in business to deliver the remarkable.”

The beauty of imaginative leaps

Advertising may be using data incorrectly (the horror!). Grace Kite argues that:

Data shouldn’t be used to dictate, instruct, or generate ideas. Coming up with good ideas is a job for people. Instead, data should be used to convert ideas into action. To demonstrate why that idea is widely relevant, to show what it can do, and to get people to act on it. That’s the superpower of good data, and advertising needs it now more than ever.

And finally…

It seems we’re increasingly embracing the imperfect, and that is not restricted to wonky veg, but so much more. 

I’m going away for a few days with my co-founders next week – always looking for ideas of things to do… maybe not this one. 

Voiceover artists display incredible skill – here is one showing 10 popular tones used in voiceover. The Bright Read… total genius. Do machine transcriptions pick this up, I wonder?