[For background on my video model work, see Storyteller Requiem.]
Business Models and Video Models
The entertainment industry relies on top-down, one-to-many models. A small number of people make a show and a lot of people (hopefully) watch it.
The nascent generative A.I. industry, in contrast, is aimed toward intense personalization. The Personal Assistant is the dominant paradigm: a user has a one on one relationship with their A.I. Existing “A.I. entertainment” exists mostly in the tawdry form of simulated relationships with robotic girlfriends.
With respect to A.I. video some people are blown away by the concept of customization: an A.I. telling stories just for me! I can make what I want!
Others focus on “creator tools”: the artists will love this. The writers will love this. They imagine a world where entertainment is still top down, but creation is assisted by AI.
But neither of these are quite right:
Full personalization is expensive. It sacrifices the gains in cost for a 1:1 relationship.
Human-assistance assumes that humans are better writers than A.I., or provide some other integral part of the process. It sacrifices the gains in cost for human employment.
What I see emerging instead is a semi-personalized process, where viewers collectively shape their entertainment in real time — as a group, or a chat, in a social way.
Assume we have an A.I. which can generate infinite animated video content on par with, say, Disney’s animated shows, for less than ten cents on the dollar.
The business model: a subscription service, A.I.T.V., which gives you access to all the shows. Think Netflix, but instead of 12$ / month, you pay 1$ / month.
Many people will be passive consumers, but there will be some new forms of interaction with the A.I. and with each other. I propose the four Cs: creation, curation, collectivization, and conversation. Let’s take a look at each:
Creation
A minority of users will work with A.I. tools to create and customize their own shows. Similar to social media today, creators will have disproportionate influence on the platform.
Curation
Power users will curate the best shows - think along the lines of Reddit moderators or Letterboxd enthusiasts. More passive viewers will “vote with their eyes” on which shows are best.
Collectivization
Fans of different shows might vote on something like “should this character die in episode 4” or “we want to see more of this character” or “here’s what should happen next”. Different groups might make different endings for their favorite show. Medium sized groups of thousands of people will coordinate with each other and the AI.
Conversation
The natural way to interface will be via chat, as with ChatGPT etc. today. Conversation *is* a form of media and the back and forth will lead to some interesting formats - think "dialogues" between AIs broadcasting media, AI influencers etc.
Social Entertainment
A.I.T.V. is not quite social media, not quite traditional entertainment -- it's something along the lines of social entertainment, mediated by A.I.
My expectation is that incumbent platforms will capture a large share of the value from these use cases. Facebook, TikTok, Youtube, will be fine. Social media may take even greater entertainment market share as A.I.s post directly to an existing chat-enabled space. But an entire universe of technology will be built in the meantime as people vie towards this future.