Shared issue
BAI with Human Responsibility at the Centre
As intelligence is increasingly attributed to machines, human judgement, responsibility and accountability risk being pushed aside.
Current status
The issue is visible; a Mission has not yet been formed.
This is a pre-establishment record explaining why the issue matters and which people could work on it. There is currently no appointed Mission lead, active Circle, verified needs study, announced calendar or open participation.
Why this issue?
Begin with the concern and the people living with it.
As intelligence is increasingly attributed to machines, model outputs can be presented as final decisions or authoritative knowledge while human judgement and accountability fade from view.
What calls our attention to this issue
- Model outputs may be presented as final decisions, authority or knowledge itself.
Why is it shared?
Human responsibility and accountability can become a shared standard in the production of knowledge and technology.
Who could carry it?
Different forms of expertise that understand and respect one another’s limits.
Professions and responsibilities
- Teacher · the place of people and knowledge
- Legal practitioner · responsibility and decision limits
- Developer · explainability and human control
- Media practitioner · visible uncertainty and sources
How can children and young people take part?
Small, age-appropriate and protected contribution paths under the care of a master or guide.
- Human-review indicator
- Uncertainty card
- Model-limits demo
A possible path if the Mission is formed
Not every issue needs to become a product.
The issue is understood first. If the need and risks call for it, the team prepares a small trial or limited production. The work may stop at the right time, remain in limited use or continue with a named person responsible for care.
- 01
Understand together
Listen to the people living with the issue, read the sources and develop a shared understanding before rushing to a solution.
- 02
Run a small and safe trial
Observe uncertainty and risk at a small scale; remain open to stopping and revising as well as continuing.
- 03
Build within clear limits
If the need is verified, form a capable team and build a limited, usable representation with someone responsible for its care.
- 04
Continue, share or stop
Suitable work may open for HAYR, remain in limited use, continue through NEMA or be withdrawn for revision.
Possible forms of work
These are not promised outcomes; they are directions that may be tried if the need is verified.
Human-review standard
Example of an explainable BAI interface
Published records related to this issue
Representations that may help you think about the issue
A published record does not mean that a Mission has begun, a team is active or its intended result has been verified. Each one is a human-made representation that must be read with its sources, limits of use and care information.
circle · vIn preparation
BAI Human Oversight Guide
A preparation record for a proposed guide on human decision, challenge and reversal in the use of models and automation; the work has not yet been published.
View the record and its limits of use →builders · vIn preparation
Builders Working Guide
A preparation record for a proposed Builder guide covering Mission choice, task sharing, SEDD and responsibility for continuing care; the work has not yet been published.
View the record and its limits of use →Share an issue or contribution
What do you know or experience about this issue?
Regular Mission participation is still being planned. You can already share the language and place in which you encounter the issue, the skills you could offer or what you would like to learn. This is not an application to an active programme.