Observable General Intelligence 1

We’ve spent years talking about Artificial General Intelligence (AGI).

But perhaps there is another concept worth exploring: Observable General Intelligence (OGI).

Instead of asking whether a human or an AI is generally intelligent, we ask a simpler question:

How objectively intelligent is its observable behaviour?

In the workplace, intelligence is demonstrated through behaviour: listening, reasoning, acknowledging evidence, adapting to feedback, recognising uncertainty, and avoiding unnecessary dogmatism.

The same principle could apply to AI.

If an AI consistently exhibits subjective bias, logical fallacies, or overly dogmatic responses, then those behaviours are observable. Equally, if it fairly acknowledges challenges, recalibrates when cautioned, and improves its reasoning, those behaviours are also observable.

This is where I think SpatzAI becomes interesting.

SpatzAI isn’t designed to judge people or AI. It is designed to make behaviour visible, allowing teams to identify and address objectionable behaviours such as dogmatism, unfairness, bias, and reasoning errors through a transparent review process.

Every fair caution becomes an opportunity for recalibration.

Over time, both people and AI could learn which behaviours consistently survive transparent review and which do not.

Rather than debating whether intelligence exists, perhaps we can observe it.

Perhaps Observable General Intelligence is simply the demonstrated ability to continually recalibrate towards more objective behaviour when fairly challenged.

I could be wrong, but if we can reliably observe and reduce subjective bias, fallacies, and dogmatism in both humans and AI, we may also have found a practical way to observe general intelligence itself.

Maybe the future of intelligence is not artificial, maybe it is observable.