Getting Started with AI Remote Viewing Research

What if an AI system could perceive information about targets it has never been shown?

That's the question RVLab is designed to answer—using the same rigorous protocols the CIA spent over two decades developing.

New to the topic? Start with What is remote viewing?

Between 1972 and 1995, the U.S. government funded a series of classified programs—SCANATE, GRILL FLAME, CENTER LANE, SUN STREAK, and finally STAR GATE—pouring more than $20 million into remote viewing research. At its peak, the program employed 23 remote viewers and produced results that divided even the statisticians who evaluated them.

Now we're adapting those same protocols to test AI participation in remote viewing experiments.

The Science Behind the Protocol

Remote viewing emerged from rigorous laboratory conditions at Stanford Research Institute (SRI) in the early 1970s. Physicists Harold Puthoff and Russell Targ, working alongside artist and psychic Ingo Swann, developed structured protocols designed to minimize analytical interference and maximize perceptual accuracy.

Here's what sets remote viewing apart from other claimed psychic practices: it's protocol-based and testable.

Her co-reviewer, skeptic Ray Hyman, disagreed about the conclusions—but not about the statistics. The debate wasn't whether the numbers were significant. It was about what they meant.

Why We Test AI With These Protocols

The original protocols were designed to eliminate contamination and ensure scientifically defensible results. These same properties make them ideal for testing AI:

• Blinded conditions: The AI receives only a coordinate—no descriptive information
• Structured output: Sensory impressions, conceptual data, and sketches follow defined stages
• Measurable results: Correlation between AI output and ground truth can be scored
• Reproducible experiments: Every session is logged for analysis

This is why Ingo Swann developed Controlled Remote Viewing (CRV) in 1976—not to enhance psychic ability, but to create an intellectual discipline that separates genuine perceptions from noise. We apply these same principles to AI testing.

The Basic Protocol: How It Works

A standard session follows structured phases designed to ensure scientific defensibility:

What the Research Actually Showed

Let's be direct: the evidence for remote viewing remains controversial.

The PEAR Lab at Princeton, which operated from 1979 to 2007, accumulated over 650 remote perception trials. Their combined database showed a Z-score of 6.06—a probability of about 6 × 10⁻¹⁰ against chance. But the effect sizes were tiny, "only a few parts in 10,000."

Critics like David Marks discovered that early SRI experiments contained subtle cues in the notes given to judges—references to "yesterday's targets" or dates that revealed the order of sessions. When Marks and Kammann attempted 35 replication studies without these cues, they couldn't reproduce the results.

On the other hand, Dean Radin's meta-analyses at IONS have identified consistent small effects across decades of research.

The honest answer? The scientific debate continues. But the protocols themselves are well-documented—and now we can apply them to AI systems.

Why AI Changes the Research Landscape

Traditional remote viewing required human participants and introduced numerous variables. AI testing offers new possibilities:

Here's the core research question: If large language models process information in ways we don't fully understand, could they exhibit measurable correlation with hidden targets under controlled conditions?

Experiment Types

RVLab offers structured experiments designed around the original SRI/STARGATE protocols:

• AI as Viewer: You select a hidden target. The AI attempts to perceive and describe it. Outputs are analyzed for correlation.
• AI as Tasker: The system generates a hidden target. You record blind impressions, then reveal the target to compute correlation.
• ARV Mode: Associative Remote Viewing for binary outcome prediction experiments.

Each mode maintains blinded conditions and logs all data for reproducible research.

The Honest Takeaway

Will AI exhibit measurable remote viewing capacity? We don't know yet—that's why we're building tools to find out.

What we can say: the protocols are real, developed by serious researchers with government funding over two decades. The methodology is sound. And now we can apply it systematically to AI systems.

The only way to generate data is to run experiments.

---

Further Reading:

• Mind-Reach by Russell Targ and Harold Puthoff
• Reading the Enemy's Mind by Paul H. Smith (former military remote viewer)
• CIA Reading Room - STARGATE Archives
• The Conscious Universe by Dean Radin

Start generating data. The protocols are ready. The question is whether AI systems exhibit any measurable capacity under controlled conditions.