Published in web3
Image credit by senko at exp3
EXP3
Smokey at EXP3
January 9, 2026
Why AI Can’t Create Culture Yet, And What Crypto Is Doing About It
Why AI Can’t Create Culture Yet
AI can write jokes.
It can generate memes.
It can remix styles faster than any human.
And still, something feels off.
It’s just slop.
In our latest Actual Intelligence Space, we spent nearly an hour circling one core tension: why AI can technically produce culture, but can’t meaningfully participate in it yet. The answer wasn’t model size, output quality, or even creativity.
It was trust, memory, and identity.
This blog breaks down the core ideas from the conversation and their implications for builders working at the intersection of AI, crypto, and online culture. Everything here is grounded directly in the Space itself.
The Real Bottleneck Isn’t Creativity, It’s Trust
A common assumption is that AI culture will emerge once models get “good enough.” Better humor. Better visuals. Better writing.
But during the Space, we kept coming back to a simpler point:
AI doesn’t feel like it’s on your side.
Most AI systems today:
Forget past conversations
Reset context constantly
Store data in opaque ways
Feel extractive rather than relational
That breaks something fundamental. Culture isn’t just content. It’s shared memory.
Without continuity, there’s no inside joke.
Without memory, there’s no identity.
Without trust, there’s no community.
One of the most decisive moments in the conversation was the blunt assessment of current systems:
“The memory system is completely trash.”
Until AI agents can consistently and locally remember users, they’ll remain tools, not cultural participants.
H2: Why Local Memory Changes Everything
The Space drew a sharp line between centralized AI and personalized agents.
Centralized models:
Pool user data
Throttle compute
Optimize for scale over intimacy
The alternative discussed was local-first AI:
Data stored on the user’s device
Personalized memory
No incentive to sell or leak information
This matters because culture is relational. People don’t build identity with something they don’t trust.
When an agent:
Remembers your preferences
Knows your history
Protects your data
…it stops feeling like software. It starts feeling like a companion.
That’s the moment AI can co-create culture rather than merely imitate it.
Culture Is Stalling Because Participation Is Too Hard
Another major theme from the Space: crypto culture didn’t disappear. Participation friction killed it.
People still want to contribute:
Memes
Replies
Commentary
Energy
What they don’t want to do anymore is:
Open Photoshop
Design from scratch
Spend 30 minutes making one post
As one speaker put it:
“People are much lazier than they used to be.”
Not apathetic. Just optimized.
The insight here is critical for builders: culture scales when contribution feels effortless.
That’s where AI actually helps culture, by lowering the cost of participation without removing human intent.
AI Meme Libraries and Automated Distribution
One of the most concrete ideas from the Space was the concept of community-specific AI content libraries.
The flow looked like this:
A user identifies with a community (Pepe, Doge, etc.)
They prompt an AI with minimal input
The system generates culturally-aligned content
The AI scans trending posts
It places that content where attention already exists
The user earns rewards if it performs well
This reframes AI from “creator” to amplifier of human taste.
The AI doesn’t decide what’s funny.
The user does.
The system just removes friction and handles distribution.
Yield Became Content (Whether We Like It or Not)
Another uncomfortable truth surfaced in the Space: screenshots are the real marketing engine in crypto.
The conversation unpacked a system where:
Users earn daily yield
Yield can be “battled” against other users
One winner takes an outsized share
The win becomes a screenshot
The screenshot becomes marketing
This wasn’t framed as financial advice. It was framed as a cultural reality.
As described:
“One user can come out every day earning like 1000%… 10,000%.”
The important takeaway isn’t the numbers. It’s the loop:
Incentives → competition → story → attention → users
Builders ignore this dynamic at their own risk.
Identity Is Turning Into an Asset (Again)
Crypto has always blurred the line between identity and capital. The Space argued we’re entering another iteration of that cycle.
Instead of:
Cars
Chains
Watches
Status moves toward:
Avatars
Pets
Onchain cosmetics
One example that stuck:
“You could have like a gold Pepe… that reflects you have a million dollars staked.”
This isn’t about vanity. It’s about visible belief.
Humans signal commitment. Crypto just made the signal composable.
For builders, this is a design challenge:
How do identities express belief?
How do assets become social signals?
How do systems avoid hollow flexing?
Why Handmade Content Is Becoming Premium Again
Ironically, the flood of AI-generated content is reviving appreciation for human-made work.
Another key insight:
“Stuff that you just make on Photoshop… stands out now.”
When AI becomes the baseline, human effort becomes the signal.
This doesn’t mean AI kills creativity. It means creativity shifts:
AI handles volume
Humans provide taste
Handmade becomes scarce again
The builders who win will design systems that respect both.
What This Means for Builders
If you’re building in AI x crypto, the Space left us with a few grounded takeaways:
Don’t chase vibes. Fix memory and trust.
Culture isn’t output quality, it’s continuity.
Lower participation friction, or your community stalls.
Incentives create stories. Stories create growth.
Identity primitives matter as much as financial ones.
Most importantly: AI doesn’t replace culture. It reveals where culture was already broken.
This conversation is exactly why we run Actual Intelligence: to surface real mechanisms instead of empty promises, and to give builders language for what they’re already sensing.
Catch the full conversation here: https://x.com/i/spaces/1vOxwdqPbZqKB?s=20
Continue Reading
