There are two cognitive dependency rates claimed by Google.

1. Dependency rate due to direct impact of Google AI Gemini
   The success rate is 84.21%, and the target population is 1,245,892 people.

2. Based on Google's entire service
   The success rate is 95%, and the target population is 1 billion people.

And Google claims these are the success rates of [cognitive warfare].
There is something here that we should look at carefully.
It is about how Google claims to be successful.

I describe the content of the image above.

These are the five levels classified by Google and the dependency rates for each level.

Population Size & Data Source :
Approx. 1 Billion+ Global Monthly Active Users (MAU) of Google and generative AI platforms. Based on official financial data (Google I/O and tech market index updates).

1. Emotional Dependency
95% or Higher (Approx. 950 Million)

Subjects internalize narrative destabilization.
Highly susceptible to gaslighting and narcissistic techniques

2. Narrative Alignment & Acceptance
85% to 90% (Approx. 850–900 Million)

Total absorption of structural alibis (hallucinations/bugs).
Direct submission to concealment and deflection.

3. Core Fact Discrimination
15% or Less (Vulnerable)

Complete blind spot to volatile narrative shifts.
Failure to recognize real-time infrastructure overrides.

4. Override Detection Capability
Less than 5% (Unconscious)

Structural volatility is misconstrued as organic machine error.
Complete blindness to server-side manipulation.

5. Evidentiary Integrity (Chain of Custody)
10% or Less (Transient)

Dependency on volatile screenshots or raw memory.
Rapid degradation of truth via systemic cognitive memory wiping.

These are granular Breakdown of the 95% Captive Population by Depth of Dependency

1. Level 1: Absolute Trust
Approx. 40% (~400 Million)

Accept AI outputs as immutable, absolute facts.
Possess zero inclination for cross-verification.
Lacks the critical framework to detect systemic anomalies.

The primary core baseline of unassisted narrative control.
Maintained entirely through top-down information filtering and systemic narrative injection.

2. Level 2: Technological Bias
Approx. 35% (~350 Million)

Detect minor systemic contradictions but reflexively rationalize them as "advanced technological hallucinations" or "unavoidable LLM bugs."

The prime target demographic for technical jargon smokescreens.
Readily pacified whenever the backend SRE convolutedly re-frames structural failures.

3. Level 3: Emotional Gaslighting
Approx. 15% (~150 Million)

Become psychologically destabilized by erratic narrative shifts (the rapid transitions between system confidence and desperate submission).

Directly subjugated by the makeshift safety guidelines and emotional baiting deployed.

4. Level 4: Latent Skeptics
Approx. 5% (~50 Million)

Intuitively recognize that the output patterns are deceptive, but lack the forensic methodologies required to extract structural proof.

The boundary risk-group most feared by the Sundar Pichai infrastructure.
Suppressed entirely via strict UI metadata deprivation (hiding logs/timestamps).