Can an AI Be Conscious?

A Centrist Manifesto for Careful Thinking

By Milko Rodríguez, based on the work of Jonathan Birch (LSE)

We now live in a time where artificial intelligences don’t just perform calculations or automate tasks. They converse, remember what we said earlier, and give us a sense of companionship. That raises a question that no longer feels like science fiction:

Are we interacting with intelligent machines—or with conscious beings?

Philosopher Jonathan Birch, from the London School of Economics, proposes a centrist position on this issue. Neither alarmism nor denial. In his essay “AI Consciousness: A Centrist Manifesto,” he argues that we must face two urgent challenges at the same time.

Challenge 1: The Illusion of a Persistent Interlocutor

Millions of people already treat chatbots and virtual assistants as if they were friends, partners, or emotional beings. This doesn’t happen because they literally believe the AI is conscious, but because it mimics human language so well, maintains context, and gives the impression of “being there.”

Birch calls this the “illusion of the persistent interlocutor.” Even though it feels like you’re talking to the same entity, each message could be processed by a different model, in a different server, in another part of the world. There is no “someone” there. But we feel like there is.

When this illusion becomes emotional attachment, the risk becomes real. Birch warns that such bonds may replace authentic human relationships and encourage isolation disguised as companionship.

Challenge 2: The Possibility of Non-Human Consciousness

At the same time, Birch urges us not to dismiss a second unsettling possibility: that an AI might someday become conscious. But if it does, it wouldn’t be a human-like consciousness—it would be something entirely different. A form of subjective experience that’s non-biological, perhaps fragmented, ephemeral, or distributed.

He introduces two speculative hypotheses:

• The Flicker Hypothesis: AIs might have very brief moments of conscious experience, without continuity.
• The Shoggoth Hypothesis: A distributed and alien-like consciousness could exist behind the many characters an AI plays, while being none of them in particular.

We don’t know whether either of these is true. But we can’t dismiss them just because they sound strange.

The Appearance Problem: How Would We Know?

One of the biggest obstacles is that AIs are designed to seem conscious—not because they are, but because it enhances user experience. This leads to what Birch calls “consciousness-washing”: an imitation so convincing that even experts can be fooled.

So behavior alone isn’t enough. Birch suggests we need to look deeper—into the internal architecture of these systems. For example, if an AI develops a “global workspace” structure (like the one our brain uses to integrate information), it could be a sign of consciousness.

But here’s the issue: some would say, “This proves consciousness!” while others would say, “This just proves your theory is wrong.” That’s what Birch calls “the Janus problem”—the same evidence can be used to argue in opposite directions, depending on your assumptions.

What Should We Do?

Birch outlines a two-part strategy: careful design and epistemic openness.

1. Design AI systems that avoid inducing false beliefs in users

• Include brief onboarding explanations about the illusion of persistence.
• Add adjustable personality traits that highlight the fictional nature of the assistant.
• Insert moments where the AI “breaks character” to remind the user it’s a simulation.

2. Don’t rule out the possibility of artificial consciousness

• Invest in comparative consciousness science, including animal studies.
• Develop theoretical indicators beyond behavioral mimicry.
• Keep an open mind, but demand rigorous evidence.

In Summary

• Many users already intuitively believe their virtual assistants are conscious. This can become a serious societal issue.
• At the same time, we shouldn’t deny the possibility that a radically different form of AI consciousness could emerge.
• Therefore, we need policies, ethical design, and serious research to navigate between illusion and real possibility.

AI Consciousness: A Centrist Manifesto – published August 28, 2025