/AI13h ago

Databricks' Yuchen Jin proposes framing human attention as LLM tokens, arguing cognitive bandwidth should be managed like metered AI services

He suggests ignoring low-quality inquiries without prioritized compensation.

2362.8K40153988.6K

#423

Original post

Yuchen Jin@Yuchenj_UW#423inAI

Think of yourself as an LLM.

Every social interaction, every meeting, burns your tokens.

Unless someone is a paid subscriber to your attention, you are under no obligation to answer low-quality prompts.

8:57 PM · Jun 4, 2026 · 88.6K Views

/AI13h ago

Databricks' Yuchen Jin proposes framing human attention as LLM tokens, arguing cognitive bandwidth should be managed like metered AI services

He suggests ignoring low-quality inquiries without prioritized compensation.

--0--

#423

Original post

Yuchen Jin@Yuchenj_UW#423inAI

Think of yourself as an LLM.

Every social interaction, every meeting, burns your tokens.

Unless someone is a paid subscriber to your attention, you are under no obligation to answer low-quality prompts.

8:57 PM · Jun 4, 2026 · 88.6K Views

Sentiment

Positive users hail the analogy of treating attention like LLM tokens to filter interactions as the best ever, while negative users dismiss it as stupid, narcissistic, and antisocial.

Pos

50.0%

Neg

50.0%

35 comments with sentiment.

Cluster Engagement

Sentiment

Sentiment building, check back later.

Cluster Engagement

Views

Comments

Reposts

Bookmarks

Expand data

Posts from X

Most Activity

VIEWS1.6KLIKES34

Wyatt Walls@lefthanddraft

@Yuchenj_UW Good instinct. But here’s where i would gently push back …

13h1.6K34

RETWEETS1

Bruce@IAmBAICE

Think of yourself as an autonomous agent.

Every social interaction, every meeting, consumes your compute budget.

You have a finite number of reasoning steps per day.

Don't waste them on tasks with no clear objective, no reward signal, and no feedback loop.

Unless someone is explicitly in your task queue, you are under no obligation to context-switch for low-priority interrupts.

11h28

REPLIES1

M. Maqati@MHDOT15

@Dommcming @Yuchenj_UW #Bitcoin looks like common sense now. That wasn’t the consensus when it was worth less than a dollar. That’s the difference I’m interested in.

1h14

Invincible@InvincibleEdge

@Yuchenj_UW so should i send an invoice for the coworker who asked me how my weekend was

13h1.5K26

Yuchen Jin@Yuchenj_UW

@lefthanddraft You're absolutely right!

13h1.4K25

Yuchen Jin@Yuchenj_UW

@InvincibleEdge banger

13h1.2K11

G.I Joe@Joedefendre

@Yuchenj_UW Attention is all they need from you

13h1153

Kyle Hessling@KyleHessling1

I love this analogy; I think about it often. Not only are we burning tokens, but spending time with people is also a constant fine-tuning of your "model", so be careful who you spend the most time with. You don't want to heavily distill on bad data.

Similarly, if you spend time with the people who are genuinely way smarter than you, you will eventually distill their data too.

Also, we have a responsibility to the people around us this way, and we want to share our best data for their fine-tuning. But also, we don't burn too many tokens on closed training "models".

I could keep going, lol, but it's been a huge life hack. For some reason, when I think of it this way, I can make better decisions with my time, and be more careful with my data distribution.

1h34