Our mission
We are not another AI platform or cloud service.
We take the hard questions AI raises about the future of humanity, and we study, experiment and test as rigorously as we can. Our aim is for humankind to reap the benefits of AI: building toward a brighter future, separating evidence from myth and hype, and staying grounded in the work of finding the AI solutions the world still needs.
What we are, and are not
We are not
Another AI platform or cloud service.
We are
People who take the hard questions AI raises about the future of humanity, and study, experiment and test as rigorously as we can.
Our aspiration
We hope to earn trust as a standards company that leads by example: honest about every result, and every result repeatable.
This is an aspiration, not a claim.
How to read this site
- Every result is run by us. No unaffiliated party has reproduced any of it yet.
- Every number comes with its public source and a short verbatim quote.
- Failures, nulls, retractions and corrections of our own work are listed with the successes.
- Outcome labels are ours; the five definitions are below.
- Most labs use small open models (SmolLM2-135M, Qwen2.5-0.5B, a pruned BitNet-2B) on a few machines we own. Read each lab's scope before drawing conclusions.
Outcome labels
Every lab carries one of five labels. The labels are ours, and each is shown with the same weight.
- Positive
The check worked as expected, within the stated scope.
9 labs - Mixed
Some parts held, some did not, or we corrected ourselves.
14 labs - Negative
The claim or approach failed, or we retracted it.
6 labs - Null
No effect found, or nothing yet to report.
3 labs - Held back
A pre-registered rule withholds the result until its cause is understood.
1 lab