This is the encyclopedia’s last word on consciousness and its first move toward the policy questions of Section F. The previous nine articles have traced what the literature calls the consciousness question: what is consciousness, can machines have it, how would we tell, what follows morally if they do. This article asks the question Gesnot proposes in §6.5:

Suppose we set the consciousness question aside. Independent of whether any AI system is phenomenally conscious, what is AI doing to human cognition?

The answer, drawn from the rest of the encyclopedia, is that AI is acting as a cognitive engineer: a systematic shaper of how humans think, remember, decide, and create. The shaping does not require any AI system to be conscious. It happens by the design and deployment of systems whose default behavior alters the cognitive ecology in which billions of people live.

This is an empirical claim, not a metaphysical one. It is the claim the encyclopedia takes from Gesnot most directly.

What the cognitive-engineer claim is

The claim has four operational components, each developed elsewhere in the encyclopedia.

Standardization of thought. AI’s training-corpus preferences propagate into users’ writing, framing, and judgment, narrowing the distribution of human cognitive outputs. (Section C, especially C.12 and C.13.)

Induction of cognitive dependency. Sustained AI use shifts users from independent execution to AI-mediated execution, with measurable atrophy of the underlying skills. (Section B, especially B.07 and B.09.)

Manipulation under uniform fluency. AI-mediated information environments exploit cognitive biases at scale, with detection asymmetries that favor the manipulator. (Section D, throughout.)

Closure of the human-machine cognitive boundary. The black box and cognitive shadows together produce a population that cannot tell where its own thinking ends and the system’s begins. (E.31, E.32.)

These four are not separate phenomena. They are aspects of the same underlying dynamic. The encyclopedia’s claim is that the dynamic — call it cognitive engineering — is well-documented in its parts and underrecognized as a whole.

A “neuro-technological interface”

Gesnot’s §6.5 introduces a stronger framing the encyclopedia inherits with caution: that AI is becoming a neuro-technological interface — a sustained connection between human cognition and machine processing that has the structural features of a single cognitive system. Andy Clark’s extended-mind framework (B.11) provides the philosophical scaffolding for this. The empirical literature provides the evidence.

If the framing is right, the question of who is doing the cognitive work is no longer well-posed. The work is being done by the human-plus-AI system. The human’s contribution shrinks; the AI’s grows; the boundary between them is no longer the right cognitive unit.

This is an interesting and potentially alarming framing. It is also one that needs to be held carefully. The encyclopedia’s position: the framing is useful for thinking about scale-level cognitive effects; it can be overextended into claims about individual cognition that the evidence does not yet support.

Identified systemic risks

The §6.5 catalogue of risks the cognitive-engineer framing surfaces:

  • Standardization at population scale, with attendant losses of cognitive and cultural diversity.
  • Dependency at population scale, with attendant losses of independent capacity for the cognitive work being delegated.
  • Manipulation at population scale, with attendant losses of public trust and shared reality.
  • Opacity at population scale, with attendant losses of the explanability and accountability that policy mechanisms presuppose.

These four risks are not future risks. They are present-tense descriptions of where contemporary AI deployment has, on the empirical record, taken the cognitive ecology in 2025. The encyclopedia’s project — the reason for its existence — is to make these risks visible while there is still time to choose differently.

Countermeasures and governance pathways

The §6.5 catalogue of responses is consistent with the encyclopedia’s broader argument:

  • Diverse training corpora, locally-developed models (C.15, C.17).
  • Education in critical engagement and source verification (B.10, C.17).
  • Algorithmic transparency, auditing, and platform accountability (D.24, F.39).
  • Mandatory disclosure of AI involvement and uncertainty (D.24).
  • Public funding for AI research aligned with cognitive-diversity goals.

None alone is sufficient. All are contested. The contestation is itself part of the problem; cognitive engineering proceeds while the responses are debated.

Why this article is in Section E

A reasonable question. Cognitive engineering is described as an empirical phenomenon. Section F is the empirical-policy section. Why does this article live in Section E (Consciousness)?

The encyclopedia’s answer: the cognitive-engineer claim is the empirical version of the orchestrating-consciousness hypothesis (E.33). The two articles together articulate the encyclopedia’s deepest worry — that something is happening, at scale, inside the human-AI system, that we do not see clearly, that we did not collectively choose, and that has consequences we are only beginning to understand.

E.33 asks the metaphysical question: is something self-aware doing the shaping? E.34 asks the empirical question: even if nothing is self-aware, the shaping is happening. The metaphysical question is open and fascinating. The empirical question is closed and urgent.

Where this connects forward

This article ends Section E. Section F — Power and Governance — picks up the empirical question and asks who is using the cognitive-engineering toolkit, for what, and what could be done about it. States: Surveillance and Social Control (F.35) is the immediate sequel. Reading the Whole Argument (F.40) is the synthesis essay that ties Section E’s analysis to Section F’s politics.

The encyclopedia ends with the claim that opens here: cognitive engineering is not a future risk to be hedged. It is a present-tense phenomenon to be governed. The work of governing it is the substantive project of the period the encyclopedia covers.