REALISM INDEX: 9.0 / 10 — Near-future credible
2019 Game · Eliza on IMDb ↗ · Official game listing ↗ · Original editorial illustration
There are no robots in Eliza. No superintelligence, no uprising, no uploaded minds. There’s a woman in a small room in Seattle, wearing glasses that feed her lines, reading them to someone who is struggling.
It’s the first entry on this index to score as near-future credible, and it earns that easily. Nothing in it needs inventing.
The proxy
Eliza is a counselling service run by a tech company called Skandha. The AI listens to the client, analyses what they say, and decides what should be said back. But the words aren’t delivered by a screen. They’re delivered by a person.
Evelyn is a proxy. She sits across from the client, reads the system’s prompts through her glasses, and speaks them aloud. She’s expected not to deviate. Her job is to provide the one thing the software can’t: a human being in the room.
This is the game’s central insight, and it’s devastating in its plausibility. The warmth is real — Evelyn is present, attentive, embodied. The judgement is not hers. The client experiences care from a person while receiving the output of a system, and the arrangement works precisely because the human makes the machine feel like something other than a machine.
The economics of scaled care
Why build it this way? Because a trained counsellor is expensive and scarce, and a proxy reading a script is neither. Skandha’s product isn’t therapy. It’s access — a version of support cheap enough to reach people who would otherwise get none.
The game refuses to make this simply villainous, and that’s where it’s sharpest. Some clients are helped. Some aren’t. Sessions tend to resolve toward standardised recommendations that serve the platform’s needs as much as the person’s. And the question the game keeps asking — is some care at scale better than real care for almost no one? — doesn’t have an easy answer, which is exactly why it’s the right question.
Evelyn knows
Evelyn helped build Eliza before she left Skandha, and she returns as a proxy knowing exactly what the system is and what it isn’t.
That makes her the most useful protagonist on this index. She isn’t discovering a hidden truth. She’s inside a system she understands, deciding whether to keep reading the lines. Every one of her choices is a real one, and the game’s multiple endings let you make them — go back to the company, go independent, walk away — without telling you which is correct.
Why it scores so high
Every element is present-tense. Conversational systems that analyse speech and propose responses exist. Mental-health platforms built on scaled, lower-cost models of care exist. The business pressure to reduce the cost of a session while increasing the number of sessions exists. Human workers performing the parts of a service that software can’t yet do convincingly exist across the industry.
The only thing Eliza invents is the particular combination, and even that feels less like speculation than a slightly early description.
It loses points only on causal consistency, where a few character arcs resolve more neatly than the rest of the game’s careful ambiguity would suggest.
The scores
| Category | Score | Why |
|---|---|---|
| Technical feasibility | 9.5 | Nothing shown requires technology that doesn’t already exist. |
| Human behaviour | 9.0 | Clients who are helped, clients who aren’t, and a proxy who knows too much. All convincing. |
| Institutional response | 9.0 | A company optimising access and cost over care, without ever becoming a cartoon. |
| Causal consistency | 8.5 | Rigorous and restrained, with a few arcs that tie off more neatly than the rest. |
Realism Index: 9.0 — Near-future credible.
Calculation: (9.5 + 9.0 + 9.0 + 8.5) ÷ 4 = 9, rounded to 9.0/10. These are editorial judgments, not measurements. Read the scoring method.
The verdict
The films on this index ask what happens when machines become people. Eliza asks what happens when people become the interface for machines — and it’s the only one describing something that could happen next year.
It’s a small game about a quiet job. It’s also the most accurate picture of deployed AI that fiction has produced.
Frequently asked questions
What is Eliza about?
A 2019 visual novel by Zachtronics about Evelyn, a former developer who works as a human proxy for an AI counselling service, reading the system’s responses aloud to clients.
Is Eliza realistic?
Highly. Every technology and business pressure it depicts already exists. It is the only entry on this index rated near-future credible.
Why is the game called Eliza?
It echoes ELIZA, the 1960s program by Joseph Weizenbaum that imitated a psychotherapist and became one of the first famous demonstrations of people attributing understanding to software.
Scope and sources: This review covers Eliza. The work listing identifies the title; American Psychological Association advisory on AI mental-health apps provides context for real-world AI risks. The interpretation and scores are ours. Reviewed September 2026.
Feasible components do not establish that an AI counselling service is clinically safe or effective.