Key Takeaways
- Sam Altman’s warning about AI-driven social media accounts echoes the Dead Internet Theory.
- The real problem isn’t bots themselves but the rise of an AI monoculture.
- A healthier internet future requires decentralized compute and diverse, community-built models.
Sam Altman recently posted on X about his concern around the proliferation of LLM-run social media accounts, essentially endorsing the “Dead Internet Theory.”
The irony is staggering.
The man whose company flooded the web with ChatGPT-generated content is now worried about AI pollution online.
X users immediately called out this hypocrisy and were right to do so. Altman helped create the exact problem he now laments. His concern, however, points to something real that’s happening across the internet.
The Dead Internet Theory suggests that most online activity comes from bots, creating a sterile wasteland where real people unknowingly interact with synthetic entities.
Altman’s post suggests he finds this scenario increasingly plausible.
What Is AI Monoculture?
The internet feels dead because we’ve created an AI monoculture. The problem isn’t that bots exist. It’s that they all sound the same.
This homogenization stems from a simple fact: Only a few companies can build and deploy large-scale language models.
OpenAI, Google, and Anthropic dominate because training advanced AI requires massive computational resources that most organizations simply cannot access.
When everyone uses the same few models, online content converges toward the same tone, style, and perspective. This Dead Internet becomes predictable and soulless because its AI-generated content lacks the diversity that makes human communication interesting.
The centralization of AI development has turned the web into an echo chamber of corporate-approved responses.
These companies optimize for engagement and risk management rather than authentic expression, creating content that feels artificial even when technically sophisticated.
The Cost of Centralized AI
Building competitive AI models requires enormous capital investment and technical expertise. This reality has concentrated AI development within a handful of Silicon Valley companies that can afford the computational costs.
The infrastructure requirements create a vicious cycle. Amazon Web Services, Google Cloud, and Microsoft Azure provide computing power that enables AI development and control access to those resources. Companies building AI must work within the constraints and priorities of these platform providers.
This dependency shapes AI development in subtle but important ways. Models get optimized for metrics that matter to cloud providers rather than end-users.
Safety guidelines reflect corporate legal concerns rather than community values. The result is AI that serves platform interests more than human creativity.
Independent developers and smaller organizations get priced out of meaningful AI innovation. They can access pre-built models through APIs, but cannot create alternatives that reflect different values or serve various communities.
Decentralization Is the Solution
The answer to AI homogenization is diversity, not elimination. We need thousands of unique, community-specific models rather than three or four corporate ones. Different communities should be able to train AI to reflect their particular knowledge, values, and communication styles.
This requires breaking the computational monopoly that enables AI centralization. Decentralized compute networks can provide the GPU power necessary for independent AI development without dependence on Big Tech infrastructure. Consider an internet where AI enhances human creativity instead of replacing it with generic content, eliminating the Dead internet and making it alive again.
When communities can train their models on distributed networks, they can create AI that amplifies rather than replaces human diversity.
An AI trained by artists would have different priorities than one trained by researchers or hobbyists. These differences would enrich online discourse rather than flattening it.
Community governance of AI development could produce models that reflect genuine human variety instead of corporate risk management. Local communities understand their own needs better than distant platform companies.
A Vision for Internet Revival
Consider an internet where AI enhances human creativity instead of replacing it with generic content. An internet where you might interact with a bot that reflects the distinct personality of a specific subculture rather than the sanitized voice of corporate communications.
Different AI models could serve various purposes and communities. Academic researchers could use models trained on scholarly literature. Artists could interact with an AI that understands creative processes. Local communities could deploy models that reflect regional knowledge and concerns.
The goal is authentic diversity rather than artificial scarcity. More AI models, built by more people, serving more communities, would create the plurality that makes online spaces interesting.
The Infrastructure Challenge
Building diverse AI ecosystems requires new infrastructure that operates independently of centralized cloud providers. Current compute monopolies prevent the innovation necessary to revive internet culture.
Decentralized networks can provide the computational resources communities need to train and deploy their AI models. These networks would operate more like public utilities than private platforms, serving community needs rather than corporate objectives.
The technical architecture exists to support distributed AI development. What’s missing is sufficient investment in decentralized alternatives. Building these networks represents both a technical and cultural challenge.
The Choice We Face
We could end up with a Dead internet dominated by a few corporate AI systems. The alternative to this is a pluralistic digital ecosystem where diverse communities can build AI that reflects their unique perspectives.
Reviving the internet means building better infrastructure for AI diversity rather than simply building better chatbots.
Developers, investors, and users who support decentralized networks can help break the monopoly that makes the internet feel dead.
About the Author
Tory Green is the Co-Founder of io.net, the world’s largest decentralized AI compute network. He’s led io.net to a $1 billion valuation and major exchange listings. His career spans investment banking at Merrill Lynch, strategy at Disney, private equity at Oaktree Capital, and leadership in multiple startups. Tory holds a BA in Economics from Stanford University and played football at West Point. He now focuses on advancing open, decentralized AI infrastructure and innovation across the AI and Web3 sectors.
