The wider conversation about machine intelligence shifted noticeably across 2024 and 2025, and it kept shifting in early 2026. That shift is the reason this program tightened its language, not the other way around.
When the ground under a conversation moves, the words that used to carry weight start meaning different things to different rooms. That is when a research program has to decide whether to ride the current vocabulary or to hold a careful frame. IamHITL, and the family it belongs to, is holding a careful frame. This piece explains what the field looks like from here, and why our two commitments, natural not artificial and humans in the loop, are the honest response to it.
What the field looks like from here
Two things are happening at the same time. Capability is climbing on several axes, and the definitions used to describe that climb are getting stretched by whoever needs them stretched. Vendors, analysts, and forecasters are all using overlapping vocabulary to mean non-overlapping things. That is not a moral failure of any one party. It is what happens when a field moves faster than its own dictionary.
A useful outside snapshot of that unsettled context comes from Themesis. Their April 2026 piece surveys how the wider landscape of "general" intelligence research repositioned itself. Class E We link it below in our own words, not theirs, because the point here is field-level context, not paraphrase.
A second Themesis piece, published later that month, walks through three industry moments that reshaped what people expect from learning systems: deep learning, transformers, and a newer architecture line. Class E Read that lineage the way we read it here. It is a pattern of scaling steps, each with real capability and real limits. Nothing in that pattern says the next step has arrived. Nothing in it says the direction of travel is understood in the same way by everyone selling into it.
Why natural, not artificial, is the honest response
If the field vocabulary is stretched, a research program can do one of two things. It can ride the current usage and count on the reader to sort out what the words mean. Or it can pick words that stay stable even while the industry ones drift, and describe its own work with those. This program does the second. The phrase General Natural Intelligence is that stable phrase. It points at the mechanism active inference actually names, prediction and update inside a living system, and it refuses to inherit the "arrived" connotation the industry vocabulary has picked up.
Our canonical sentence stays the same on every site in the family. UNI is a working hypothesis on an attainable path toward General Natural Intelligence: a natural, active-inference approach whose evidence is growing, evidence-classed, and tested in the open. Do not take the claim on faith. Test the build, inspect the gates, and help us find where it fails.
Why human-in-the-loop is the other half of the frame
Language matters, and so does control. If the field is producing tools whose behavior is uneven and whose vocabulary is contested, the safe deployment posture is not to hand a tool the last word on anything that matters. It is to keep a human in the decision, wired into the loop, with visible seams around the tool's outputs and a clear place to intervene. That is what "human-in-the-loop" names on this site. It is not a warning label attached to a finished product. It is a design commitment: the loop is drawn with a person in it from the start, and the tool sits inside it, not around it. Class C
Field context is why this matters right now. A moment when definitions drift is exactly the moment when quiet delegation to tools accumulates fastest. The counter-move is to make the loop visible, to name who decides, to keep receipts, and to leave room to disagree with the tool.
How to read anything else on this site
Read every claim here as a calibrated signal, not a verdict. Where a claim rests on someone else's writing, we link the writing and describe it in our own voice, tagged Class E. Where a claim is unverified, we tag it Class U and hold it lightly. Where a claim rests on how our own systems are built or configured, we tag it Class C. The vocabulary of this program is stable on purpose, so it can carry the field context without pretending to have settled it.
The frame itself lives in the natural, not artificial explainer. The partnership posture lives in partner, not replacement. The place where humans and teams practice the loop together is the workshop.
Why we say natural, not artificial
The plain-language explainer for the phrase General Natural Intelligence and what we deliberately do not claim.
The posturePartner, not replacement
How this program treats machine tools inside a loop that keeps the human wired in.
The workshopWork with us directly
Two days of practicing the loop with humans in the room. Applications and details.
The mathUNI, the science front
The technical side of the same program, with citations, benchmarks, and open gates.
Evidence classes used in this post: Class C configuration and integration, Class E expert citation, Class U unverified. Full method and taxonomy at the transparency page.