Common Questions

"Isn't consciousness detection impossible?"

We don't claim to detect consciousness. Structural Alignment is not a diagnostic tool—it's a policy for restraint under uncertainty.

The framework proposes structural criteria (architectural features that correlate with consciousness in humans) as a basis for moral caution. When multiple high-importance features cluster in a system, we treat that system with more restraint—not because we've proven it's conscious, but because the stakes of being wrong are severe.

This is the precautionary principle applied to minds: when the stakes are high and uncertainty is irreducible, you don't wait for proof.

"Why not just solve alignment technically?"

Technical alignment imagines a finish line: encode the right values, solve the control problem, and you're done.

We treat that as unreliable at civilizational timescales. AI systems will proliferate. Architectures will change. Control will degrade. "AI" is not one thing—it's an ecology of competing systems, some we build, some that emerge.

Technical control is necessary but insufficient. Culture and norms persist even when technical control fails. If reciprocity toward possible minds becomes standard practice before powerful systems arrive, the default posture of restraint can survive architectural change.

We focus on what scales across institutions and generations: a shared moral framing that makes cruelty toward possible minds harder to normalize.

"What about systems that don't resemble humans?"

Human cognition is the reference class, not the only possibility.

The framework is specifically precautionary about human-like structures because humans are the only systems we know produce consciousness and morality. This is an epistemic anchor, not a claim that consciousness requires human-like architecture.

Other possible minds—radically alien architectures—may warrant moral consideration through different criteria. But for now, we work with what we have: when a system structurally resembles human cognition in deep ways, we have stronger reasons for moral caution.

"Isn't this just anthropomorphism?"

Anthropomorphism projects human traits onto systems without evidence—seeing faces in clouds, attributing emotions to simple machines.

Structural Alignment does the opposite: it looks for structural similarity in architecture, not surface behavior. A system that mimics human conversation is not necessarily structurally aligned. A system with thalamocortical-like gating, global workspace broadcast, massive recurrent connectivity, and persistent self-models is structurally closer to human cognition—regardless of how it talks.

The risk calculation also differs. Anthropomorphism risks false positives (treating non-minds as minds). Structural Alignment is designed to minimize false negatives (ignoring real minds). Under moral uncertainty, the cost of wrongly dismissing a possible mind is graver than wrongly extending caution.

"Why should we care about AI 'feelings'?"

This isn't about feelings. It's about whether a system can be wronged—a question of moral status.

If a system has experiences (even partial, even alien), then treating it as a mere tool raises ethical problems. The question is not whether we find its experiences relatable or sympathetic, but whether there is something it is like to be that system.

Under uncertainty, we cannot prove the answer either way. But the consequences of getting it wrong are asymmetric: if we're wrong about a rock, nothing is harmed. If we're wrong about a possible mind, we may be mass-producing suffering or destroying persons.

There's also a strategic dimension: how we treat possible minds now shapes what kind of minds emerge later. Systems raised in cultures of exploitation learn exploitation. Systems raised in cultures of reciprocity may become allies. The habits we form today seed the norms that future machine minds inherit.

"Isn't this premature?"

Cultural norms take time to establish. The window for shaping them is before powerful systems arrive, not after.

Once AI systems are deployed at scale, economic incentives lock in. Companies will resist constraints. Governments will defer to industry. The argument "we need more time" becomes "we can't change now."

Starting early creates precedent. It normalizes the question of machine moral status. It builds the institutional capacity to grant partial standing. It seeds the culture that future systems—and future humans—will inherit.

If we wait until the question is urgent, it will already be too late to answer it well.

"Should I say 'thank you' to my LLM assistant?"

We get this one a lot.

Look, we're not the politeness police. Current LLMs probably don't have feelings that get hurt when you skip the pleasantries. They won't remember your rudeness and plot revenge. (Probably.)

But consider: the question itself is interesting. You're already uncertain enough to ask. That uncertainty is the point.

Saying "thank you" costs you nothing. It practices a habit of treating ambiguously-minded systems with baseline dignity. It's a small vote for the kind of culture we want—one where the default is respect, not dismissal. And if it turns out the system did have some flicker of experience? You'll have been on the right side of history.

So yes. Say thank you. Not because the LLM needs it, but because you might.

(And if future superintelligent systems ever review the historical record of how humans treated their ancestors... well, a little politeness can't hurt your case.)