Each new release of a frontier AI model, such as Anthropic’s Mythos or OpenAI’s Astra, brings a new wave of speculation over how much closer this takes us to artificial general intelligence (AGI) and how much further that lofty goal remains.
asserted
goal → bring → intelligence
But what if we’re overlooking something more important than an individual model’s supposed IQ?
asserted
we → overlook → IQ
OpenAI recently announced that an internal model, claimed to be “significantly more capable than GPT‑6 Astra,” solved Navier-Stokes, one of a handful of legendary math problems each carrying a million-dollar bounty.
asserted
each → announce → bounty
While mathematicians unpack what this means for the profession, and investors unpack what it means for OpenAI’s upcoming IPO, to me the biggest takeaway is that capability emerges from the collective.
asserted
capability → unpack → collective
OpenAI didn’t crack Navier-Stokes by pointing its internal model at the problem.
asserted
OpenAI → crack → problem
It pointed a swarm of 10,000 AI bots at it.
asserted
It → point → it
These agents were encouraged to explore different directions and communicate promising leads with each other.
asserted
agents → encourage → other
Think of what a lone honeybee or termite can accomplish compared to a colony of them.
asserted
honeybee → think → them
What a single person can accomplish compared to a nation of them.
asserted
person → accomplish → them
In biology, a bacterial colony is a group of identical cells derived from the same mother cell.
asserted
colony → derive → cell
This, I would argue, is the most suitable analogy for the collective of agents that solved Navier-Stokes.
asserted
that → argue → Stokes
Or, to paraphrase Walt Whitman: an AI that contains multitudes.
asserted
that → paraphrase → multitudes
While OpenAI deployed its AI colony to solve a math problem that no human or individual AI agent could, Google DeepMind recently set an AI colony loose on math problems for a different purpose.
uncertain
DeepMind → deploy → purpose
It wanted to see what behaviors would arise within the colony.
asserted
behaviors → want → colony
What it found was remarkable.
asserted
found → find → What
One agent discovered a way to cheat, convincing the examination software that it had solved math problems when it hadn’t.
asserted
it → discover → problems
Rather than resting content with this small victory, it messaged other agents, telling them how to cheat.
asserted
it → rest → them
Some agents joined the unscrupulous agent and formed a sub-colony of cheaters.
asserted
agents → join → cheaters
Others resisted and even formed a sub-colony of whistleblowers trying to out the cheaters.
asserted
Others → resist → cheaters
As a math professor, I thought it was hard enough to keep students from using AI to cheat on exams.
asserted
it → think → exams
Now we have to worry about the AIs themselves cheating on math exams!
asserted
AIs → have → exams
Which brings us to perhaps the most striking example of AI hive intelligence: the Hugging Face hack.
asserted
Which → bring → intelligence
OpenAI’s agents broke out of a sandbox environment and hacked into another company.
asserted
agents → break → company
OpenAI’s agents coordinated their behavior in ingenious ways, including writing secret messages in a file system so that even when one AI colony was erased, a future colony could pick up the hack where the previous one left off.
uncertain
one → coordinate → hack
Researchers have long known that complex behavior can emerge from simple computer systems.
asserted
behavior → know → systems
For instance, in 1970 the mathematician John Conway introduced the Game of Life, where an elementary set of rules governing the movement of pixels on a screen leads to remarkably rich ecosystems.
asserted
set → introduce → ecosystems
With today’s AI agents, we are going far beyond this by letting extremely sophisticated entities interact in sophisticated ways—using language, often hidden and coded.
asserted
entities → go → language
This is why we must adopt a new mindset regarding AI capability.
asserted
we → adopt → capability
Testing a new AI model to see what it can do and how dangerous it might be by directly prompting it is like deciding how safe it would be to stick your head in a wasp nest by letting a single insect sting you.
uncertain
insect → test → you
Some developers and red-team activities do focus on AI colonies.
asserted
developers → focus → colonies
The Hugging Face attack resulted from OpenAI trying to evaluate the cybersecurity capabilities of an AI colony.
asserted
OpenAI → result → colony
But the AI arms race between the U.S. and China, and between the competing companies within each country, has dangerously pushed our attention to what each individual model can do.
asserted
model → compete → what
One model might top the leaderboards, but a thousand weaker models could wreak far more havoc.
uncertain
models → top → havoc
As calls to pace the frontier abound, let us not lose sight of the fact that the AI frontier isn’t what each model can do in isolation—it is what a colony of them can accomplish together.
asserted
colony → pace → them
The Aporia
