![[Matter AI.png]]
##### We said a while back this year that the practice of politics would be defined by “AI” - even if the term itself sows its own confusion. [The mainstream press is on the beat.](https://www.wsj.com/politics/elections/ai-deepfakes-are-getting-weirder-and-harder-to-spot-in-the-midterms-88b4f7ad?st=HVbcem&reflink=desktopwebshare_permalink) We’ll pile on with some context.
##### What we learned from Spencer Pratt in LA
1. **[This is one cool application of AI](https://www.youtube.com/watch?v=ZJIBBUyVIq0) to make entertaining political content.** Notice, it’s not bot-slop. And the viewer is in on the joke from the beginning. They’re not trying to fool you. It's fun. Your eye keeps looking for another Christopher Nolan reference and other fun easter eggs. Even if, in this case, the politics was wrong.
- Important note: underemployed studio hands with right-wing sympathies can do a hell of a lot with AI and some time on their hands. We bet this was cut in the off-hours of someone tasked with the next Star Wars slop-fest. Or, even more likely, someone laid off in the near-depression that is the current Hollywood production economy. Either Way, Do Not Try this at Home. It’s not amateur work.
- **Highly watchable political content is not the same as a good piece of advertising.**
2. **Pratt's team didn't make the stuff that made him most famous.** Again, most campaigns won’t be able to say that - running a populist campaign in the capital of the world’s film industry helps a lot. But we expect to see more crowd-sourced creative as insurgent candidates take on crumbling blue-city infrastructure in particular. (Wait for the Mamdani re-elect)
3. **Reality TV is good training for candidate skills in the 21st century.** Expect more of this too; for good or for ill.
##### What the Pratt campaign may portend
**Attentional plays and electoral plays are still far apart - sort of.**
We talk about the merging of attention and persuasion all the time. But attention “on the internet” writ large vs the attention of a narrow slice of your voting population are very, very different things - tactically and strategically. And campaigns confuse the two at their peril.
The correlation between attention and political strength is probably most true on the activist left, where internet virality is more tied to cash in the bank: Platner, Mamdani, Talarico, the DSA types in the recent NY primaries.
But here’s an interesting related point from Kyle Tharp, in the Upper West Side’s Congressional district:
Attention does not equal votes, in anything remotely resembling equal measure. We so rarely make that distinction tactically or strategically.
Getting lots and lots of people across the country (world) to chuckle and watch a 2:30 second ad on social media is a fundamentally different task than getting the right 50,000 people in a specific geography (e.g. LA City, or your state Senate district) to digest your message and change their behavior because of it.
Funny or not, 2:30 is too long for a campaign ad.
**The hardest job in advertising is getting people to believe you. AI is making that job more difficult.**
The negative ad of the future is about situating the subject in context, not dropping ad hominems on the subject.
Manufacturing thin-reed claims and slinging them out of context at a candidate with low-name ID will have a place. But better if the viewer first understands the urgency of the problem visually. We know the land of Gotham in Batman’s universe isn’t a nice place. We don’t need to be told. If a viewer has that establishing context, the hits land harder.
## First, Get them to believe you… if you can.
The practical, everyday downside to the upside of the Pratt crowdsourcing: nothing in this moment in time will make a voter disregard your argument faster and better than the suspicion that you made this ad with "AI altered” images or words - AI in quotations because we use computational animation in numerous, appropriate, sensible applications across our production process.
[We made our own AI botslop checklist.](https://bullhornthematter.com/Feature+Notes/On+Political+Botslop#The+Botslop+Checklist)
Example: family grocers used to be everywhere. If you needed to shoot a spot that illustrated high grocery prices, you might find a friendly store owner willing to let your candidate into the meat aisle and make his case.
Today; good luck getting that permission from Aldi or Wal-Mart.
So why not simulate a grocery aisle in a see-it-hear-it cut? The viewer gets the point without being asked to take a visual leap of faith.
The incongruent break of faith and attention – “should I really believe that?” - is a fatal crash for any piece of advertising.
AI just makes it easier for teenage boys to race drunk along the cliffs.
[Here's a great summary of the problem of slop and believability, with solid citations](https://medium.com/@sauravsinghsisodiya/the-internet-is-filling-up-with-ai-content-and-readers-are-pulling-away-bd5c1f609bc0). It references written digital content... we suspect the conclusions apply to video at least equally; if not more so. Nut Graph:
> **The trust paradox nobody's talking about.**
> In blind tests, 84% of readers couldn't tell the difference between AI writing and human writing, according to a Firewire Digital study. So technically, most people can't spot it. But 52% disengage when they think content is AI, *even if they're wrong about it*.
> **That's not a detection problem. That's a trust problem.**
> Once readers suspect something was machine-generated, they're done. It doesn't matter if it was actually written by a brilliant human. The suspicion is enough. Which means the content flood isn't just lowering the quality of what's online. It's lowering the quality of how readers approach everything online.
*“The suspicion is enough.”*
^ Heed those words before approving any attack ads this cycle. **If you’re looking for a believability checklist, [this is a good start](https://bullhornthematter.com/Feature+Notes/Every+Image+Matters+\(The+Power+of+the+Frame\)).**
### Spots Worth Watching for Trends’ sake
1. [NRSC shaking up the visual context of a good negative hit](https://x.com/NRSC/status/2064497773379231796?s=20) - ***The best negative ads don't look like other negative ads***
We've mentioned before how, every decade or so now, the Willy Horton ad re-emerges.
The spot worked then. So long as the audience doesn't know how stale it is; the spot works now.
Advertising doesn't have much in the way of plagiarism laws. The best stuff is almost always imitation and adaptation - especially from the ever-so-distant past.
We're guessing NRSC picked this Dot-Matrix era treatment because they want to appeal to women 45+; which is where Maine will be fought. Nostalgia plays.
And we're guessing they made it in-house... it's torturously over-written, and should be condensed to :30 seconds for air. It's a broadcast-ready concept.
This spot would be much less effective with big, slash-and-flash typefaces.
Or simply too many cuts. The way most negative ads look these days.
The same simple visual framework maintained throughout helps the narrative break through. The eye doesn't have to work to understand what's going on, so the ear can focus on the cheeky, clean copy and the well executed VO. (AI?)
Our take: trim the fat in the script, get it to :30, and get it on the 6pm news in Bangor. Throwback, indeed.
2. **Never forget [TV is a visual medium](https://bullhornthematter.com/Feature+Notes/Humans+Learn+Visually)!**
[Check out this spot](https://bullhorncomms.slack.com/files/UUXJYR8CR/F0B9Y9BP006/swfzyokok2xucs0os9dabocsrjs.mp4) from progressive city council candidate running on Capitol Hill in DC. She lost.
It could be a Republican spot: highlight the pretty blonde white lady being victimized. And as a viewer, you're not so sure she isn't a candidate herself for a moment...
Until we cut to the endorsement and the baggy (red!) T-shirt, the quirky glasses, and the Asian lady who, complete with the slanty logo, is a progressive Democrat.
It's a tidy little ad in concept and execution… but the broader feel seems off. Is she a progressive? Or is she the candidate of white victimization politics?
Attempts to code left while talking right is going to be a new hotness in urban politics, we predict, [as failed governance models are rejected by angry voters](https://www.governing.com/politics/brandon-johnsons-woes) from the [left and right.](https://www.slowboring.com/p/progressives-need-to-reckon-with?utm_source=substack&utm_medium=email)
Coding left while talking right? Not sure where that might apply. If you can find a good example of a pink-haired Asian lady talking America F#ck Yeah Bald Eagles and Gun-Porn and winning for it; please send our way.