惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

博客园 - 三生石上(FineUI控件)
D
Docker
GbyAI
GbyAI
宝玉的分享
宝玉的分享
Jina AI
Jina AI
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
Vercel News
Vercel News
博客园_首页
Recent Announcements
Recent Announcements
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
Hugging Face - Blog
Hugging Face - Blog
腾讯CDC
S
SegmentFault 最新的问题
Microsoft Security Blog
Microsoft Security Blog
酷 壳 – CoolShell
酷 壳 – CoolShell
美团技术团队
V
V2EX
WordPress大学
WordPress大学
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
V
Visual Studio Blog
IT之家
IT之家
Apple Machine Learning Research
Apple Machine Learning Research
T
Tailwind CSS Blog
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
Universal Basic Tokens: The Future of Technological Capit...
Adrien Cossa · 2026-04-23 · via DEV Community

What if, instead of Universal Basic Income, the powers that be hand us AI tokens? Those who manage to spin those tokens into something commercially viable will survive. The rest will still get their tokens, but they won't be able to trade them for food or shelter—and they will be easy to blame. In our increasingly technocratic society, this feels much closer to reality. Why give people money when you can give them tokens? It introduces an extra layer of control, a new gateway for exclusion.

I am actually writing this by hand (well, it doesn't mean I won't do a quick AI proofreading pass later) because I am staring down the barrel of my quota. Time to step back and share some thoughts. Why not?

Claude Code Usage

I have a Claude Code Max 20x subscription—basically the ceiling before you're forced into pay-as-you-go—but I only have 3% of my limit left. This isn't just the rolling 5-hour window limit; this is the hard weekly stop. It's manageable this time (although I still have work to do, and honestly, I dread doing it manually), but if you hit this wall after just two or three days of the week... good luck.

I try to conserve my tokens, but this is inevitable. Am I wasteful? Yes. Could I optimize? Sure, and I've tried. But the reality is that you can never have enough tokens. Having an AI implement a feature is not intrinsically demanding. It's the context. It’s loading the massive playbooks containing your best practices, business logic, and operational guidelines. It’s retrieving data from third-party APIs, parsing PDFs, scanning multiple repositories, digesting thousands of lines of error logs, and debugging end-to-end tests via screenshot analysis.

The more you automate, the more context you need, and the faster your quota burns. My attention is already fragmented across five or six concurrent sessions. If I want to stay focused, I can't be bothered to constantly switch models or manually compact prompts to save pennies. Yes, there are tools to help manage this (like writing a custom statusline hook), but there is no silver bullet. To borrow a metaphor: asking a developer to ration tokens is like asking a painter to use as little paint as possible while trying to express her art.

For software engineers, agentic AI has become more than a drug; it's basic infrastructure. We need it every single day, just like we need the internet. We are still doing software engineering, but AI has naturally promoted us all to software architects and engineering managers. The tools do us a lot of good, but the system around them is taking its toll.

Just look at the quota anxiety it creates.

  • Scenario one: It’s 12:30 PM, and you’ve used 37% of your 5-hour window, which resets at 1:00 PM. Time for a break? No. You don't want to let that remaining quota go to waste, because whatever you skip now might be impossible to compute in four hours when you're bursting at the seams trying to fix a massive suite of e2e tests.

  • Scenario two: The 5-hour windows aren't your main problem. Just working normally through the day is enough to hit that hard weekly limit. But what happens when you don't hit it? If you happen to have surplus tokens near the end of the billing cycle (unlike me today), you stay up late working just to make sure you use them all by the end of the week. Any unused token is a resource thrown in the garbage that you might terribly miss in just a few days.

Pick your poison.

Either you work in a big company where you are encouraged to consume thousands of dollars a month—because it means you fully embrace the trend and are super-performing, though that comes with its own kind of stress and sleeplessness since infinite tokens mean infinite expectations. Or you pay for the extra usage without blinking because it yields a high ROI. Otherwise, you are like most people, trying to get through the month with your flat-rate plan, knowing that you will eventually hit the cap and have to fall back on the API and/or supplement it with extra usage—you just don't know under what circumstances that shoe will drop.

It is fair to pay for a high-quality service, despite the occasional outages. But let's be honest about the promise of AI: it was supposed to boost our productivity and take the weight off our shoulders. Instead, we are working longer hours and dealing with more stress than ever. If you thought AI was a tool you could just set and forget while you relax, you misunderstood the assignment—and you might not keep your job for long.

This relentless pace and constant token-juggling is ultimately what brings us back to the broader picture. Universal Basic Income might happen just to keep the economic machine running, but don't be surprised if that "income" comes in the form of tokens to capitalize on.