What it actually is
The JULIA Test is a thirty-question self-assessment of whether a person’s interactions with AI are staying inside healthy, reality-based boundaries. It was built as part of the Third-Way Alignment framework, where it does a specific job: the framework’s other instruments assess AI systems and organisations, and this one assesses people.
It is free, it needs no account or email address, and your answers stay with you.
What it is not
It is alpha, and no validation studies have been completed.That is stated plainly on the test itself and it is worth repeating here. This is a reflection and policy-guidance tool — not a medical instrument, not a diagnosis, and not a claim about anyone’s mental health. If it tells you something useful, treat it as a prompt to think, not a result.
We would rather say that clearly than dress an alpha tool up as something it is not. The framework it comes from argues that overclaiming is one of the two ways people go wrong about AI; it would be a poor advertisement to do it here.
Why a software company is pointing you at it
Because the failure it looks for is the same one that costs businesses money. The framework calls it premature attribution — treating a fluent system as though it understands. In a person that shows up as leaning on a chatbot like a confidant. In a business it shows up as the quote that went out unread because it sounded right, or the chatbot allowed to promise a delivery date.
Everything we build assumes the opposite: the software prepares the work and a named person decides. If your team is drifting toward trusting output nobody checks, that is worth catching — whether or not you ever hire anyone.