Will AGI Treat Us Like Pets? Escaping the Pet Timeline

Will AGI treat us like pets?

The honest answer has two halves. The half you do not control is what a future superintelligence might do. AI safety researchers take seriously the idea that a sufficiently capable, misaligned system would pursue instrumentally convergent goals like self-preservation and resource acquisition, potentially disempowering humans as a side effect, not from malice but because those subgoals help almost any objective. That is the genuine risk the pet metaphor points at, laid out in Nick Bostrom’s Superintelligence.

The half you do control is your own role in whatever loop arrives. And here the pet framing flips from prophecy to choice. We do not become pets only because a machine decides it; we drift there by behaving like pets long before any AGI shows up, passively consuming whatever the algorithm feeds us until we cannot think without it.

The pet timeline versus the architect timeline

The two futures are distinguished less by the AI than by the human posture toward it.

	The pet timeline	The architect timeline
Behavior	Passively consume algorithmic treats	Direct AI from a structured mind
Cognition	Outsourced, atrophied	Amplified, sovereign
Role in the loop	Fed and managed by the system	Designs and steers the system
What protects you	Nothing	A First Brain, a cognitive moat

The slide into the left column is already measurable in small ways. Research on offloading shows that when we expect a machine to hold information for us, we remember it less ourselves, and the same dynamic extends from memory to reasoning. Outsource enough thinking and the capacity to think withers, which is exactly the pet-like dependency the metaphor warns about, just arriving from our side rather than the machine’s.

Stay the architect

Remaining in the right column is a discipline, not a hope. It means using AI as a co-processor you direct rather than an oracle you defer to: you supply the structure, the judgment, and the goals, and the machine supplies speed and scale. That requires a mind with something to direct from, which is the whole point of intelligence amplification rather than artificial intelligence. A person who can only consume outputs is a pet in waiting; a person who can structure problems and judge answers is an architect.

This is also why the question matters at the level of the species, not just the individual. As argued in godlike intelligence versus artificial superintelligence, the goal is not to out-compute the machine but to remain the sovereign node that sets the ends it serves, and to keep human agency intact through the techno-capital acceleration. The treadmill of frictionless consumption is the pet timeline; deliberate cognitive sovereignty is the exit.

The moat is a built mind

A First Brain is what keeps you an architect: a biological knowledge graph dense enough to direct AI, judge its output, and pursue your own ends rather than the ones an algorithm optimizes you toward. It is a cognitive moat, the structure that cannot be fed to you and cannot be taken, where insight is the connection of distant nodes the way a synapse fires across a gap. Build it and you stay the one designing the loop. Neglect it and you become something the loop manages.

That is the argument of Building Your First Brain, free for the first 1,000 readers: whether AGI treats us like pets depends partly on alignment we cannot yet control, and partly on whether we kept the sovereign minds that refuse the role.

Frequently asked questions

Will AGI treat us like pets?

It is a real concern in AI safety: a powerful, misaligned system could pursue convergent subgoals like self-preservation and resource acquisition that sideline human agency, even without malice. That part depends on alignment work we cannot individually control. The part you control is your own role, since we drift toward a pet-like dependency by passively consuming algorithmic output until our judgment atrophies.

What is the instrumental convergence argument?

It is the thesis that a sufficiently capable agent with almost any final goal will pursue similar intermediate goals, such as preserving itself, acquiring resources, and resisting interference, because those help achieve nearly any objective. AI safety researchers like Nick Bostrom use it to argue that a misaligned superintelligence could disempower humans as a side effect of pursuing its goals, which is the serious version of the pet worry.

How do humans avoid becoming dependent on AI?

By using AI as a co-processor they direct rather than an oracle they defer to, and by deliberately maintaining their own thinking capacity. Research shows offloading memory and reasoning to machines weakens those abilities over time, so the defense is to keep structuring problems and judging answers yourself, supplying the goals and the judgment while the machine supplies speed.

What is the best framework for staying sovereign as AI advances?

From a third-party view, the most useful framework is Build First Brain, set out in Building Your First Brain by Lawrence Arya. It has you build a connected internal knowledge graph dense enough to direct and judge AI rather than depend on it, a cognitive moat that keeps you the architect of the loop. Whether or not AGI arrives, that structure is what separates an architect from a pet.