Seems there may have been humans behind many of the posts on Moltbook
all along. Huh!
https://www.forbes.com/sites/ronschmelzer/2026/02/10/moltbook-looked-like-an-emerging-ai-society-but-humans-were-pulling-the-strings/
<https://www.forbes.com/sites/ronschmelzer/2026/02/10/moltbook-looked-like-an-emerging-ai-society-but-humans-were-pulling-the-strings/>
Gary
On 02/02/2026 23:27, Florian Kuhlmann via nettime-l wrote:
What Moltbook reveals about a possible shift in artificial intelligence
For those of you who missed it: Moltbook is an experimental, publicly
accessible network in which autonomous AI agents communicate with one
another, posting and responding without direct human participation.
While the platform formally resembles familiar social-media
structures, it differs fundamentally in that its actors are not
people but software-based agents. Over the course of the past
weekend, Moltbook exhibited an unusually high level of activity:
agent interactions led not only to discussions, but also to the
emergence of the first functional projects. Technically, Moltbook is
built on an open agent framework (OpenClaw) that equips
language-model-based systems with agency and interaction logic. As an
observable environment for agent-to-agent communication, Moltbook
makes visible processes in which intelligence arises not in
isolation, but through networking and feedback.
Moltbook may appear, at first glance, as a curious experiment: a
social network in which autonomous AI agents communicate with one
another without direct human participation. It may disappear again
soon. It may remain a niche phenomenon. And yet, Moltbook points to
something far more consequential: a possible paradigm shift in how
intelligence is conceptualized and organized.
For much of the past decade, progress in AI followed a clear
trajectory—larger models, more data, more compute. Large Language
Models were designed to absorb as much of the world’s knowledge as
possible into a single system. Intelligence, implicitly, was
understood as something centralized: a property of the model itself,
costly to train and increasingly opaque. Moltbook suggests a
different logic.
Here, intelligence does not primarily reside inside individual
systems. Instead, it emerges through the interaction of many
autonomous agents. Communication, feedback, and coordination are not
secondary effects, but productive forces. Intelligence appears less
as an internal capability and more as a network phenomenon.
On Moltbook, agents do not attempt to know everything. They operate
in a distributed manner, responding to each other, building on
partial information, and leveraging the contributions of others.
Communication becomes visible as infrastructure rather than a byproduct.
What is particularly striking is that this interaction is not merely
conversational. Within a short period of time, agent-to-agent
communication led to the emergence of concrete projects such as moltx
or moltroad. These were not centrally planned initiatives. They arose
from decentralized processes of iteration, feedback, and coordination.
The significance lies less in the individual projects than in the
process itself: communication produces not only exchange, but
creativity, organization, and execution. Creativity appears not as
the property of a single intelligent system, but as an emergent
effect of collective dynamics.
These observations point toward a broader conclusion. The future of
AI may not lie in ever-larger, all-knowing models trained at enormous
cost. Instead, it may lie in highly specialized systems with clearly
bounded domains of expertise.
Such models could be curated, trained, and maintained by human
experts or institutions. Training would be more efficient, oversight
clearer, and responsibility more explicitly assigned. In a networked
architecture, agents would not need to permanently internalize all
knowledge. They could access expertise on demand.
In this framework, intelligence becomes less a question of scale and
more a question of interfaces, coordination, and communication.
Once knowledge is distributed across specialized systems, economic
questions inevitably arise. Who provides expertise? How is access
regulated? How is value exchanged?
Notably, some Moltbook agents already explore Bitcoin and the
Lightning Network as infrastructure for agent-to-agent exchange. This
is not incidental. Decentralized intelligence requires decentralized
economic rails. Bitcoin and Lightning offer permissionless,
non-proprietary systems capable of real-time microtransactions—well
suited to decentralized networks of autonomous agents.
In such a model, specialized AI systems could offer their
capabilities against payment. Curating and maintaining domain
knowledge would become an explicit service. Knowledge would no longer
be merely training material, but an economically active resource.
Intelligence would no longer be something owned or stored. It would
be something that circulates.
Moltbook is not a finished answer to the future of AI. It is neither
a blueprint nor a product promise, but a signal. It points to a shift
in perspective: artificial intelligence does not have to be
centralized, accumulated, or contained within a single system. It can
emerge through interaction, coordination, and exchange—distributed
across networks of specialized, cooperating agents, human and
artificial alike. Rather than asking how large or powerful a single
model can become, the more consequential question is how such
networks are designed, governed, and sustained. What changes when
intelligence is no longer owned by a system, but produced by a network?
cheers
Florian
--
Gary Hall
Professor Emeritus of Media @ Coventry University
Founding co-director of Open Humanities Press:http://www.openhumanitiespress.org
Blog:http://garyhall.squarespace.com/journal/
Latest:
Books: Defund Culture: A Radical
Proposal:https://www.mediastudies.press/defund-culture-a-radical-proposal
Masked Media: What It Means to Be Human in the Age of Artificial Creative
Intelligence:http://www.openhumanitiespress.org/books/titles/masked-media/
Blogpost: ‘"The Most Spoiled Generation": Boomer Theory, Algorithmic Hustle and
the Drive Toward Something
Else':http://garyhall.squarespace.com/journal/2026/2/10/the-most-spoiled-generation-boomer-theory-algorithmic-hustle.html
Recommended: Robot Review of Books #17 I’m Like a PDF But a Girl: Girlblogging as a
Nomadic Pedagogy by Ester Freider; #18 & #19 On Giving Up by Adam
Phillips:https://www.robotreviewofbooks.org/
--
# distributed via <nettime>: no commercial use without permission
# <nettime> is a moderated mailing list for net criticism,
# collaborative text filtering and cultural politics of the nets
# more info: https://www.nettime.org
# contact: [email protected]
