A language model answering a query does not experience the moment when it decides a human is no longer needed to write the paragraph, moderate the thread, or staff the help desk. It simply produces the output. The human role is not deleted by an act of will; it erodes because nothing in the system registers that the role was there. Alexis Ohanian, who helped build one of the largest repositories of human conversation on the internet, has been watching that erosion from an unusual vantage point — as the cofounder of Reddit, as an early investor in Anthropic, and now as a venture capitalist who says building social media apps is ‘not cool anymore.
The group chat as the last human room
Ohanian told WIRED senior correspondent Lauren Goode at the inaugural WIRED World Fair that he expects ‘some better version of the group chat’ to become the next form of user-generated content. [1] The remark is easy to read as nostalgia. It is better understood as a concession: the open, public, post-anything internet that Reddit once represented has been so thoroughly colonized by machine output that the only reliably human space left is the small, closed, invitation-only conversation. The role of the anonymous public poster — the person whose contribution made the platform valuable — has been quietly superseded by the bot that can produce the same contribution faster and at no cost. What remains for people is the room with a door.
A resignation and a covered tattoo
His relationship with the site he built is not simple. Ohanian resigned from Reddit’s board in 2020, after the killing of George Floyd, to open a seat for a Black director and to press the company on content moderation. The platform had become known for hosting violent and harmful rhetoric. He describes recently learning of someone who had once been so proud of the site that he had its logo tattooed on his body — and who had since had it covered up. Ohanian does not defend the site. He says he does not blame the man. That small admission carries the larger story: the human judgment that once made a platform’s reputation has been displaced, and the archive it produced is now more valuable as training data than as a community.
Where the value went
The economics tell the same story in plainer terms. Reddit licensed its archive of human conversations to companies including Google and OpenAI to train their models. Ohanian’s summary is blunt: platforms like his became very valuable to large language models. The human authorship that gave the archive its worth is now raw material, and the people who wrote it are not the customers of the resulting product. When Reddit’s shares plunged 29 percent after the company warned that Google’s AI-powered search was hurting referral traffic, the shift became visible: the machine reads the conversation, summarizes it, and answers the question, so the reader never arrives at the page where the conversation lives. The human contributor’s role — to attract a visitor — is performed instead by a system that answers in the contributor’s place.
The commercial incentive to falsify
Ohanian is a proponent of the Dead Internet Theory, the idea that much online activity is generated by bots rather than people. He does not treat this as a technical accident. The internet has become so manipulated, in his account, because it is so incredibly valuable to do so. Where the reward for producing authentic human signal is small and the reward for manufacturing synthetic signal is large, the population of human voices thins out. He worries that the commercialization of conversation has filled it with what he calls “bullshit.” [1] The judgment about what is genuine — once a reader’s job, then a moderator’s — has been handed to systems that cannot reliably make it and have no incentive to.
Interfaces built for agents
The displacement Ohanian describes goes beyond content. ‘The internet that I grew up on … that was really built for humans first; user interfaces and experiences are going to go away,’ he said, ‘because more of the internet is going to get created for these agents to use, [rather] than for people. He argues that the user interfaces and experiences built for people are going to go away, because more of the internet will be created for agents to use than for humans. That is a structural claim about who the internet is for. A page designed for a software agent does not need to be legible, pleasant, or even meant to be read by a person. The human as the assumed audience — the reader, the clicker, the commenter — stops being the design constraint. The role of the user, in the most literal sense, becomes optional.
Proving there is a person present
Reddit’s response has been to require users to sign in, a move aimed at blocking third-party scraping and holding on to engagement. Ohanian declines to say whether he supports it, but he names the underlying problem precisely: there has to be provable humanity on the internet without being creepy. [1] That sentence contains the whole difficulty. Verifying that a participant is a person requires collecting evidence about that person, and the collection is itself a surveillance act. The skill of simply being recognized as human — once free, automatic, unremarkable — now requires credentials, and the credentials cost privacy.
The case he still makes
Ohanian is an early Anthropic investor and does not conceal his optimism. He now runs the venture capital firm Seven Seven Six, which manages over $1.3 billion in assets. He is especially eager for medical breakthroughs, and he insists there is still a very real chance the technology causes a lot of harm. His position is that we have to be sober about how we build and what we do, while he still cannot help but see all of the things that can get so demonstrably better. That is not a contradiction so much as a description of where the argument actually sits: the same capability that writes the paragraph also reads the scan. The question is not whether the machine can do the work. It is which human role disappears when it does, and whether anyone notices the vacancy.
The consequence nobody states
The final implication runs further than most discussions are willing to go. If the internet is increasingly built for agents, if the valuable archive is human conversation sold as training material, if the interfaces vanish and the group chat is the last room with people in it, then the human contribution is not being replaced task by task. It is being reclassified — from participant to raw material. The person who once made the platform worth visiting becomes the substrate the platform is trained on, and the platform’s own success depends on needing fewer of them over time. Ohanian’s sober optimism is the honest position available. The unstated consequence is that a system can keep working beautifully while the role it was built around quietly empties out, and no model will ever flag the moment it happens.