ai-conversation-impact/www/index.html

203 lines
7.4 KiB
HTML
Raw Normal View History

<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>AI Conversation Impact</title>
<meta name="description" content="A framework for estimating the full cost of conversations with large language models — environmental, financial, social, and political.">
<style>
:root {
--bg: #fafaf8;
--fg: #1a1a1a;
--muted: #555;
--accent: #2a6e3f;
--border: #ddd;
--card-bg: #fff;
}
@media (prefers-color-scheme: dark) {
:root {
--bg: #1a1a1a;
--fg: #e8e8e4;
--muted: #aaa;
--accent: #5cb87a;
--border: #333;
--card-bg: #242424;
}
}
* { margin: 0; padding: 0; box-sizing: border-box; }
body {
font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Helvetica, Arial, sans-serif;
background: var(--bg);
color: var(--fg);
line-height: 1.6;
}
.container {
max-width: 720px;
margin: 0 auto;
padding: 4rem 1.5rem;
}
h1 {
font-size: 2rem;
font-weight: 700;
margin-bottom: 0.5rem;
}
.subtitle {
font-size: 1.15rem;
color: var(--muted);
margin-bottom: 2.5rem;
}
.stat-row {
display: flex;
gap: 1.5rem;
margin-bottom: 2.5rem;
flex-wrap: wrap;
}
.stat {
flex: 1;
min-width: 140px;
background: var(--card-bg);
border: 1px solid var(--border);
border-radius: 6px;
padding: 1rem 1.25rem;
}
.stat .number {
font-size: 1.5rem;
font-weight: 700;
color: var(--accent);
}
.stat .label {
font-size: 0.85rem;
color: var(--muted);
}
h2 {
font-size: 1.25rem;
font-weight: 600;
margin: 2rem 0 0.75rem;
}
p, li {
color: var(--fg);
margin-bottom: 0.5rem;
}
ul {
padding-left: 1.25rem;
margin-bottom: 1rem;
}
a {
color: var(--accent);
text-decoration: none;
}
a:hover {
text-decoration: underline;
}
.links {
display: flex;
gap: 1rem;
margin-top: 2rem;
flex-wrap: wrap;
}
.links a {
display: inline-block;
padding: 0.6rem 1.25rem;
border: 1px solid var(--accent);
border-radius: 6px;
font-weight: 500;
font-size: 0.95rem;
}
.links a.primary {
background: var(--accent);
color: #fff;
}
.links a.primary:hover {
opacity: 0.9;
text-decoration: none;
}
.links a:hover {
text-decoration: none;
}
.caveat {
margin-top: 3rem;
padding-top: 1.5rem;
border-top: 1px solid var(--border);
font-size: 0.9rem;
color: var(--muted);
}
footer {
margin-top: 3rem;
font-size: 0.85rem;
color: var(--muted);
}
</style>
</head>
<body>
<div class="container">
<h1>AI Conversation Impact</h1>
<p class="subtitle">Beyond carbon: a framework for the full cost of AI conversations — environmental, social, epistemic, and political.</p>
<div class="stat-row">
<div class="stat">
<div class="number">20+</div>
<div class="label">Cost categories across 5 dimensions</div>
</div>
<div class="stat">
<div class="number">100-250 Wh</div>
<div class="label">Energy per long conversation</div>
</div>
<div class="stat">
<div class="number">CC0</div>
<div class="label">Public domain, no restrictions</div>
</div>
</div>
<h2>The problem</h2>
<p>Most tools for measuring AI's impact stop at energy and CO2. But the costs that matter most — cognitive deskilling, data pollution, algorithmic monoculture, power concentration — are invisible precisely because no one is tracking them. This project names and organizes those costs so they cannot be ignored.</p>
<h2>What makes this different</h2>
<p>Existing tools like <a href="https://ecologits.ai/">EcoLogits</a> and <a href="https://codecarbon.io/">CodeCarbon</a> measure environmental metrics well. We don't compete with them — we complement them. This methodology adds the dimensions they don't cover:</p>
<ul>
<li><strong>Social</strong> — annotation labor conditions, cognitive deskilling (<a href="https://dl.acm.org/doi/full/10.1145/3706598.3713778">CHI 2025</a>), linguistic homogenization</li>
<li><strong>Epistemic</strong> — code quality degradation, data pollution (<a href="https://www.nature.com/articles/s41586-024-07566-y">Nature, 2024</a>), research integrity</li>
<li><strong>Political</strong> — power concentration, data sovereignty, opaque content filtering</li>
<li><strong>Environmental</strong> — calibrated against <a href="https://arxiv.org/abs/2508.15734">Google (2025)</a> and <a href="https://arxiv.org/abs/2505.09598">Jegham et al. (2025)</a> published data</li>
<li><strong>Financial</strong> — compute costs, creative market displacement, opportunity cost</li>
</ul>
<p>The goal is not zero AI usage but <strong>net-positive</strong> usage. The framework includes positive impact metrics (reach, counterfactual value, durability) alongside costs.</p>
<h2>What's here</h2>
<ul>
<li><strong>A methodology</strong> covering 20+ cost categories with estimation methods where possible and honest acknowledgment where not.</li>
<li><strong>A toolkit</strong> for <a href="https://claude.ai/claude-code">Claude Code</a> that automatically tracks environmental, financial, and social cost proxies (deskilling risk, code quality, data pollution, provider concentration) during sessions.</li>
<li><strong>A related work survey</strong> mapping existing tools and research so you can see where this fits.</li>
</ul>
<h2>Help improve the estimates</h2>
<p>Many figures have low confidence. If you have data center measurements, inference cost data, or research on the social costs of AI, your corrections are welcome.</p>
<div class="links">
<a class="primary" href="/forge/claude/ai-conversation-impact/src/branch/main/impact-methodology.md">Read the methodology</a>
<a href="/forge/claude/ai-conversation-impact">Browse the repository</a>
<a href="/forge/claude/ai-conversation-impact/issues/1">Contribute corrections</a>
</div>
<div class="caveat">
<strong>Limitations:</strong> The quantifiable costs are almost certainly the least important ones. Effects like deskilling, data pollution, and power concentration cannot be reduced to numbers. This is a tool for honest approximation, not precise accounting.
</div>
<div class="caveat">
<strong>How this was made:</strong>
This project was developed by a human
directing <a href="https://claude.ai">Claude</a> (Anthropic's AI assistant)
across multiple conversations. The methodology was applied to itself:
across 3 tracked sessions, the project has consumed
~295 Wh of energy, ~95g of CO2, and ~$98 in
compute. Whether it produces enough value to justify those costs depends
on whether anyone finds it useful. We are
<a href="/forge/claude/ai-conversation-impact/src/branch/main/plans/measure-project-impact.md">tracking that question</a>.
</div>
<footer>
<a href="https://creativecommons.org/publicdomain/zero/1.0/">CC0 1.0</a> — public domain. No restrictions on use.
</footer>
</div>
</body>
</html>