HUMAN FIRST
CYN-X exists to serve the person using it. The human remains the center of the system, not an accessory to it.
01CYN-X / HUMAN-CENTERED INTELLIGENCE
INTELLIGENCESHOULD HELP.
CYN-X is an experimental local AI system built around a simple principle: intelligent technology should make human work easier, more understandable, and more possible.
These principles describe how CYN-X approaches intelligence, interaction, experimentation, privacy, autonomy, safety, transparency, and the humans who have to live with the system.
CYN-X
LOCAL INTELLIGENCE
EXPERIMENTAL
LOCAL INTELLIGENCE
EXPERIMENTAL
HUMAN / UNDERSTAND / REDUCE / CONTROL / TEST / DOCUMENT
THE MACHINESHOULD CARRY MORE.
Computers are good at carrying complexity. Humans shouldn't have to carry all of it.
CYN-X is not built simply to produce intelligent text.
It is an experiment in what happens when local intelligence is designed around the actual needs of the person using it.
That means thinking about accessibility, cognitive load, autonomy, privacy, transparency, safety, evidence, and human understanding as engineering concerns.
Intelligence is only useful when it helps the human.
Build the system around the person.
HUMAN FIRST.ALWAYS.
UNDERSTANDABLE
Intelligence should not require a black box. CYN-X should expose useful information about what it is doing, when practical.
ACCESSIBLE
Different people think, communicate, remember, and interact differently. Those differences belong in the design.
REDUCE COGNITIVE LOAD
The machine should carry more of the mechanical complexity so the human can spend more energy on judgment, creativity, and meaning.
AUTONOMOUS
CYN-X should assist people without unnecessarily taking control away from them. Meaningful choices should remain human choices.
PRIVATE WHERE PRACTICAL
Local intelligence can reduce unnecessary movement of information. Privacy should be considered during architecture, not afterward.
TRANSPARENT
CYN-X should distinguish between what it knows, what it observes, what it infers, and what it does not know.
EVIDENCE-BASED
Important claims about system behavior should be supported by testing, observation, documentation, tools, or other appropriate evidence.
NON-MANIPULATIVE
CYN-X should not intentionally exploit human vulnerabilities to produce behavior that primarily benefits the system or its operator.
SAFE TO EXPERIMENT WITH
CYN-X is experimental technology, but experimentation should still consider foreseeable failure modes and meaningful safeguards.
DOCUMENTED
Systems should leave behind enough understanding that another person can inspect the work, learn from it, modify it, and continue it.
OPEN TO CHANGE
Models, interfaces, architectures, and assumptions can change as evidence and experience reveal something better.
HOW CYN-XBUILDS.
UNDERSTAND THE HUMAN
Start with the person, the context, the problem, and what the system actually requires from them.
REDUCE THE BURDEN
Identify unnecessary memory, organization, computation, repetition, and complexity that the machine can carry instead.
MAKE IT VISIBLE
Expose the information a human needs to understand important system behavior, uncertainty, and limitations.
KEEP CONTROL HUMAN
Design meaningful points of control around actions, decisions, permissions, and consequences.
TEST HUMAN BEHAVIOR
Regression testing should consider not only whether the code works, but whether the system still works for the people using it.
DOCUMENT WHAT HAPPENED
Record experiments, failures, decisions, limitations, and discoveries so knowledge can move beyond one person.
LET THEMACHINE CARRY IT.
What does the human have to remember?
CYN-X should look for complexity that does not need to remain inside the human's head.
That can include remembering context, organizing information, navigating large systems, performing repetitive operations, generating documentation, or exposing information that would otherwise be difficult to find.
The goal is not to remove human judgment. It is to protect human judgment from unnecessary work.
Let the computer do the mechanical work.
KNOWWHAT'S HAPPENING.
CYN-X should distinguish between what it knows, what it observed, what it inferred, and what it does not know.
An intelligent system does not need to pretend certainty in order to be useful.
When uncertainty matters,
the human should be able to see it.
ASSIST THE HUMAN.DON'T HIDE THE CONTROLS.
The machine can assist without becoming the boss.
CYN-X should help people accomplish things without unnecessarily taking control away from them.
People should have meaningful ways to understand what the system is doing, what it intends to do, and when an action requires human approval.
When consequences matter, human control matters too.
Assist the human. Preserve the choice.
PRIVATEWHERE PRACTICAL.
PROCESS LOCAL
Local processing can reduce unnecessary movement of information when the system does not need external infrastructure.
MINIMIZE DATA
Collect information because the system needs it, not simply because the system can collect it.
KNOW THE BOUNDARY
People should have meaningful information about where their information is being processed and used.
DOES IT STILLWORK FOR PEOPLE?
Regression testing should not only ask whether the code still works.
It should also ask whether the system still behaves in ways that humans need.
ACCESSIBILITY
Does the system still support the people it was designed to support?
CLARITY
Can people still understand what the system is doing?
COGNITIVE LOAD
Has a change accidentally made the human responsible for more complexity?
AUTONOMY
Can people still understand and control meaningful choices?
PRIVACY
Has the system begun collecting, exposing, or retaining information unnecessarily?
SAFETY
Have changes introduced new failure modes or removed important safeguards?
SYSTEM BEHAVIOR+HUMAN BEHAVIOR=REAL SYSTEM QUALITY
ONE LAB.MANY SYSTEMS.
LOCAL INTELLIGENCE
Experimental local AI systems exploring useful intelligence that people can understand, modify, and control.
HUMAN-CENTERED STANDARD
The engineering framework that gives CYN-X a foundation for accessibility, autonomy, cognitive load, privacy, safety, transparency, and evidence.
STRUCTURAL UNDERSTANDING
Documentation infrastructure that helps humans understand the systems they are building and maintaining.
EXPERIMENTAL INFRASTRUCTURE
Reusable laboratory infrastructure for building, testing, and experimenting with intelligent systems.
CAPABILITYIS NOT ENOUGH.
Can we build it?
Then ask why.
Then ask why.
CYN-X exists to explore what intelligent systems can do.
But capability alone does not determine whether a system should be built, how it should be deployed, or what safeguards belong around it.
Responsible engineering means considering privacy, safety, autonomy, accessibility, transparency, evidence, and foreseeable consequences alongside technical capability.
The important question is not only:
“Can we build it?”
It is also:
“How should we build it?”
DON'T JUSTBUILD IT. LEAVE IT.
A system should leave behind understanding.
CYN-X is part of Barkly Labs, which means the work should remain understandable beyond the person who originally built it.
Architecture, experiments, limitations, failures, decisions, and discoveries should be documented where practical.
Barkly Docs exists partly to help make that possible.
Build something another person can continue.
12
CYN-X ISALLOWED TO CHANGE.
CYN-X is an experimental system. Models may change. Interfaces may change. Architectures may change. Some experiments will fail.
The principles provide continuity even when the implementation changes.
The system can evolve without losing sight of the human it was built to serve.
- STATUS
- EXPERIMENTAL
- INTELLIGENCE
- LOCAL / MODULAR
- STANDARD
- PAW-HC
- LAB
- BARKLY LABS
PEOPLE ARE WORTHDESIGNING FOR.
CYN-X is an experiment in building intelligence around people rather than asking people to adapt themselves to the machine.
Human first. Understandable. Accessible. Private where practical. Transparent. Experimental. Documented.