👽Getting Ready for the Great Weirdening
When comms are going gonzo, it's awesome: Context Collapse #204
TLDR: Weirdness = End of monoculture / Democratization of creation / Technological assists
Everything in media and marketing and advertising and PR is weird right now. You don’t need me to tell you that.
But it’s *not* just AI making it weird. Blaming everything on generative AI is an easy culprit and an easy out.
So let’s talk about why stuff is so damn weird.
More goodness after the jump.
1. RIP Monoculture
I’m not the first person to go on and on about why the monoculture is dead (For starter web reading: Kyle Chayka, Simon Reynolds).
The monoculture is different from pop culture. Pop culture is evolving and glorious, but the old monoculture of three broadcast networks and a few dozen cable networks and a widely read local newspaper accompanied by a few hundred national publications is over.
Once upon a time, being a consumer of a niche form like heavy metal, Detroit techno music, British sitcoms in the USA or specialty romance novels meant going to specialty stores, building personal ties with other people into the same forms, or otherwise medium- and high-friction dedication to seeking stuff out.
Now we have YouTube, Netflix, TikTok, Facebook and the whole damn internet. Remove the friction, enable people to seek out their niches and the monoculture erodes really, really quick.
This is what we saw in 2000-2015 or so. Which brings us to…
This post is sponsored by Ungerleider Works.
This is Neal Ungerleider’s newsletter. Neal runs Ungerleider Works. Ungerleider Works helps clients with podcasts, email newsletters, ebooks, video scripts and more.
Click on the link below to schedule a free consultation:
Ungerleider Works runs on Notion. Use our link for a Notion discount.
Ungerleider Works uses The Futur’s Operating System to manage projects, clients, and sales. Get Operating System here.
2. Democratization of Creation
Smartphones are fucking magic. Even if you have a cheaper or older smartphone, you have a perfectly serviceable professional-grade still camera, video camera, home computer and professional audio-visual editing suite at your fingertips. And if you have a high-end smartphone, you have some of the best creative equipment in the world.
This is universal. It makes creating media easy and makes creating professional media easy. Any random person with a good smartphone can take on the best Hollywood has to offer and has access to global audiences.
New world.
3. Technological Assists
Building on that last one… let’s go back to generative AI, our current neat new toy/moral panic/existential quandary of 2023.
Any person with a smartphone or laptop now has access to ChatGPT, Microsoft Bing AI, Midjourney, Dall-E or Google’s Bard, which are intensely powerful creative tools that can turn just about any idea or hypothetical (if properly expressed in a format a chatbot interface can parse) into a finished or semi-finished creative product.
Awesome!
But even before that, we had foreign language captions in videos automatically translated. Someone making videos for a Brazilian audience, say, instantly has access to global viewers who don’t know a bit of Portuguese. Or grammarbots that give the masses access to professional editors. Or tools like Adobe and Canva that massively lower the learning curve for graphic design.
Of course that will make things weird. Of course.
Reader Mail:
From friend of the newsletter Andy Gryc of ThirdLaw autotech marketing in response to my last newsletter on generative AI tools:
I agree on Twitter. I don’t agree on ChatGPT and thought I’d share a couple reasons why.
I think “AI as a tool” is accurate, but that means that one person guiding AIs will be taking the place of several creatives. The exponential rate of improvement and model sizes has made ChatGPT go from cute toy to powerful tool in the span of a year.
I agree that labeling an AI as a “human-like” thinker is incorrect. They aren’t sentient. But, on the other end of the spectrum, it’s also not right to think of them as just “letter predictors” either. That ignores the phenomenon of emergent behavior, where things that the model wasn’t explicitly trained on become absorbed into its model. Papers that study emergent behavior on generative models show task performance curves with a lot of hockey sticks as the model suddenly “gets it”. Tasks the model couldn’t do at all can, after a reasonable amount of additional data/processing is added, now be executed with skill. This “aha” moment is more akin to human learning models. As the models get more sophisticated, it becomes a lot harder to extrapolate their performance, making it very hard to predict the next generation performance – other than to say that they will be remarkably better after every new release.
Lastly, yes inputting AI prompts is admittedly a bit fiddly right now to get the results you want. You know what source of data they really need? How to make prompts! Millions of people are happily doing that right now, giving them tons of data on how to appropriately create prompts and what makes for a good prompt vs bad one. I fully expect the next big announcement will be a director-level AI that generates prompts for another lower but broader AI, removing even the need for that level of human creativity.
Just welcome our robot overlords, Neal. They’ll take it easier on you that way.
What do you think? Let us know.