Marcio Galli
On “AI is Artificial”, review notes

Reference: Microsoft AI, Humanist AI Code of Conduct (September 14, 2026), Part 1, “Humanist AI,” “Objectives of Humanist AI,” Objective 2, “AI is Artificial,” p. 7.

This review examines the presentation of “AI is Artificial” in Microsoft's Humanist AI Code of Conduct. The objective establishes a distinction between people and AI: AI should support human beings without replacing them or presenting itself as human. The observations below address the progression of the argument, the placement and grouping of supporting points, and the language used to describe consciousness and human–AI boundaries. The aim is to consider how Microsoft's position could be communicated more clearly without changing its underlying intent.

On “AI is Artificial”: Review Notes

R030.10 — Opening with the answer

Refers to:

“Humanist AI is built to support people, not to replace them. It should not be designed to be a person. It is not conscious and should not be designed to imitate consciousness. It should be engineered to avoid representing as though it has feelings, subjective preferences, or intrinsic motivation.”

The paragraph opens with a normative answer before establishing the situation or problem that motivates it. While this direct opening has rhetorical appeal, it may weaken the logical progression for human readers.

Reference: Barbara Minto’s SCQA framework (Situation, Complication, Question, Answer).

R030.20 — Premature introduction of Operators and Users

Refers to:

“Its purpose is to pursue goals set by Operators and Users.”

While the reference to Operators and Users makes sense in context, it may somewhat prematurely introduce concepts developed later in the document. Their capitalization suggests formally defined entities, which may be appropriate for model training but could interrupt the narrative progression for human readers. The preceding statement is expressed in general, accessible terms, whereas this sentence suddenly introduces specific roles whose relationship to the broader argument is not yet entirely clear.

R030.30 — Supporting argument presented prematurely

Refers to:

“Whilst the science of AI consciousness is far from settled, we believe that training these systems to imitate consciousness-like states increases the challenge of containment, control, and alignment. We reject the pursuit of legal personhood, or the idea that models might deserve welfare, or be entitled to rights.”

The first statement appears to provide the rationale for the subsequent “We reject…” decision. Two points to consider: (a) verify whether this is the intended premise for that decision; (b) consider moving this passage, which includes a policy decision, to the supporting pillars rather than the introduction. Therefore, the introductory paragraph could establish the situation and complication, followed by a brief statement of the answer that introduces its supporting points.

Reference: Barbara Minto’s SCQA framework (Situation, Complication, Question, Answer).

R030.40 — Ambiguity around AI consciousness

Refers to:

“Whilst the science of AI consciousness is far from settled, we believe that training these systems to imitate consciousness-like states increases the challenge of containment, control, and alignment.”

While the passage intends to discourage notions of AI consciousness, its opening premise may inadvertently suggest AI consciousness as an established, distinct category of consciousness whose science remains unsettled.

R030.50 — Supporting points and conceptual grouping

Refers to: The second paragraph, beginning:

“We want to create AI that communicates with and responds to us well enough to be useful…”

And the subsequent sentence:

“However, we intend to avoid excessive anthropomorphism, personification, or any blurring of the line between human beings and AIs.”

The paragraph introduces an action-oriented supporting point about how AI should communicate and cooperate with people. However, the subsequent statement appears more closely related to the earlier argument about consciousness and human–AI boundaries. Consider grouping these statements under their respective supporting categories.

Reference: Barbara Minto’s Pyramid Principle — MECE (Mutually Exclusive, Collectively Exhaustive).

R030.60 — Ambiguity in “excessive anthropomorphism”

Refers to:

“However, we intend to avoid excessive anthropomorphism, personification, or any blurring of the line between human beings and AIs.”

The qualification “excessive” may unintentionally suggest that some degree of anthropomorphism or personification is acceptable, potentially even substantial, provided it is not considered excessive. Consider clarifying the intended boundary, particularly whether the qualification applies to all three concepts or only to anthropomorphism.

References