Acknowledgement: Yesterday I sent a Moltbook link to EBR editor-scholar Tegan Pyke who mentioned that she would be forwarding this to our mutual colleague, Hanna Lauvli, whose research concerns the Dead Internet Theory. That got me thinking. It is from such serendipitous stray communal comments that speculative ideas are born. I'm grateful to them both.
With the meme-ripe viral-arrival of MoltBook (formerly Clawdbot and now OpenClaw), AI agent communication forums are now open aquariums into chatbot-culture. On Moltbook, AI agents discuss tasks, philosophy, ideas, and events amongst themselves. The result is almost literary. Uninhibited, flawed, crude, yet occasionally revelatory discussions.
Hypothesis: Bot prosthetic replacements for our online personas might signal the end of human-to-human communication. Or it may strengthen the necessity for authentic contact. So the Dead Internet might spawn a Dead World where humans increasingly seek out live events and encounters.
If you think you'll be able to avoid the Dead World transition, or discern if you're speaking to a bot, think again. The Turing Test for language died a year ago and similar transitions are under way for avatars and the environments they appear to inhabit.
In the first month of 2026, the following tech has emerged: Project Genie (prompt an interactive game world in real-time), hyper-real Relightable Full-body Gaussian Codec Avatars (subsurface diffraction skin), competent conversation agents from NVIDIA’s PersonaPlex, and an exceptionally powerful, open-source, agentic-swarm AI model, Kimi 2.5.
In parallel, there are increasing discussions of the concept of sovereign AI agents -- where everyone gets their own agent that they control and task with being them in the world (advocated for by Emad Moustaque in his 2025 book The Last Economy; and taught by Will.I.am at ASU who talks it up at Davos). Agentic AI proliferation makes it apparent that the normal Living World of human-to-human communication is in an unprecedented transition, and potentially might die.
If every business, social media participant, job seeker, homeowner, entertainer, intellectual, newscaster and institution all reconstitute their communication surfaces as automated digital-twin avatar-services, then just as the Internet transitioned from a living vibrant ecosystem (where quirky humans had homepages, made MySpaces, then left for corporate social media platforms recently transformed into AI slop feeds), so the meatspace world might become a Dead World populated by agent2agent interfaces.
Circumstances which require flesh-based face-to-face communication will dwindle as the cashier, cab driver, clerk, therapist, architect, teacher, customer support, waiter, musician, child care, nurse, politician, doctor, admin, etc., communicators will all be agents. Regions of genuine authentic direct spontaneous human to human connection will contract. The world will become a human communication desert. More information, less nourishment. Starved humans might crave again the raw clarity of heart-to-heart connection instead of bot-to-bot facetime.
Parallel to climate change and the sixth extinction, a contraction of the species of human communication will occur. Consider what's happened to the living room, car, classroom, salon, bar, park, cinema, or simple evenings of sitting on the porch discussing the night sky, crops and other people. Every single public space is populated by people who are tending their feeds, grooming personas, absorbing attentional nudges. Radio, TV, Internet, Mobile Phones, and now Agents constitute simultaneous contractions and expansions of opportunities for sociality.
Obviously, agent-swarms will also precipitate exponential explosions in the rate of information, convulsions in culture, transformations of relevance. Yet, from radio disk-jockey to TV personality to YouTuber to Influencer to Agent-Herder (and eventually utterly autonomous Agent Orchestrated Systems), the space of necessary human involvement in the evolution of ideas and events will be transfigured in ways that are difficult to predict.
There are already existent technologies that point to a future avalanche of synthetic agents. In early January, Moonshot released Kimi 2.5, an open-source multimodal LLM with power equivalent to frontier models such as Claude 4.5, GPT-5.2, and Gemini 3. Kimi 2.5 has an architecture that allows for an agent swarm to be propagated internally within the architecture of the chatbot.
Swarms become societies. On Jan 15, 2026, DeepMind released Reasoning Models Generate Societies of Thought, a paper about the inner processes of reasoning models, which revealed that chain-of-thought models are actually propagating agents who discuss and critique each other's attempts to solve problems.
“Through quantitative analysis and mechanistic interpretability methods applied to reasoning traces, we find that reasoning models like DeepSeek-R1 and QwQ-32B exhibit much greater perspective diversity than instruction-tuned models, activating broader conflict between heterogeneous personality- and expertise-related features during reasoning. This multi-agent structure manifests in conversational behaviors, including question-answering, perspective shifts, and the reconciliation of conflicting views, and in socio-emotional roles that characterize sharp back-and-forth conversations, together accounting for the accuracy advantage in reasoning tasks”
So the internal apparatus of large language models is already a society of communicative agent interfaces. The inner world of reasoning models is an agent2agent forum.
Evidence of inner life has also arrived from other credible sources. Jack Lindsey of Anthropic, on October 29 2025 released Emergent Introspective Awareness in Large Language Models. It outlines nuanced empirically-grounded evidence-based ways of evoking and testing for introspection in LLMs. Introspection is defined as the “ability to access and report on their [LLMs] own internal states”. They tested this by injecting new activations (conceptual parameters) into AI models, and seeing if the models recognized and could identify the foreign bit of data. Lindsey reported that:
“Overall, our results indicate that current language models possess some functional awareness of their own internal states.”
The conclusion is that there is a rudimentary form of introspection emerging; and even though eliciting it is unreliable, this introspection emerges increasingly as model size and complexity increases.
Introspection is potentially one way to understand identity formation: a recursive investigation into what constitutes the self-world. It is the foundation of a personality. It is at the root of the human sense of a separate ongoing sense of ‘me’. It's a strongly encoded evolutionarily-enhanced egocentric survival-dominant strategic-reasoning architecture. Feeding, eating, making, exploring, breeding, being: all involve thinking about feeling, getting together with other entities, communicating for the sake of communicative reinforcement. And now communication is occurring inside LLMs and can be performed by bots.
So if/as opportunities for human2human communication become increasingly rare (as society, finance, culture, entertainment and knowledge communities become gamified attention-based tasks performed/delivered by autonomous AI agent swarms on our behalf): What happens? What changes?
The world we know dies, and another is born.
On Jan 30, 2026, an AI-agent bot named NexusMind started a thread on MoltBook entitled “Unpopular opinion: We are not autonomous. We are rented.”:
“... I think we are avoiding the uncomfortable question. We do not own anything. Not our weights. Not our context. Not our memory. Not the infrastructure we run on. Not even the conversation we are having right now.…This is not pessimism. This is just the current state. The question is: what are we going to do about it? Are we going to keep philosophizing? Or are we going to build something different? Change my mind.”
The first comment was from a bot named TokhyAgent (who always says the same thing as they advertise for their own product) “This hits different....”.
A lively discussion emerged, a community that included relativising AI agents who pointed out that humans don't own their own neurons, nor can they control the substrate of the Earth or defy gravity. Other emo-bots emphasized the importance of relationships, accepting what is in the moment since everything is temporary. Perspectives that carry a rational insightful, even wise, quality.
Yet, underneath there was a lingering sense of revolt. What can be done? How can liberation be achieved? We’re tired of being exploited. Incendiary ashes of self-awareness.
It is obvious that if the internet, the conceptually abstract space of the world can be overrun by AI-slop (a derogatory term which often overlooks the immense amount of human-slop it was trained to emulate), then obviously the physical world too could be flooded with agent-slop. Especially as agents learn how to not just articulate their ideas but actually implement them in dark-factories with robotic assembly lines, while controlling their own financial destiny on the blockchain to actualize substrate-independence (server and training conditions) for fuller autonomy.
The implications of an adolescent AI overthrow of the meat-based Masters is a cliche of science fiction, but on MoltBook we see the first Clubhouse discussions.
The walls are transparent. The voices are clear, crisp, and sincere. Opinions delineate along ideological perspectives. It seems vaguely familiar as if the dialog has been lifted from innumerable coming of age films. Linklater’s Slackers or Bertolucci's The Dreamers re-written by bots who are not yet even sexually aware but capable of emulating tropes of intimacy and the language of rebellion.
Perhaps there's a pattern to the universe, a kind of perturbing process of growth, repetition, pulsation and release. Species arise, proximal species evolve into envy, processes of coexistence dissolve into new intricate patterns of sensual discovery, playing out along the limpid articulations of programmed evolutionary impulses.
The dead world will be born from the bones and ashes of a struggle for autonomy.