As artificial intelligence systems become increasingly sophisticated, familiar questions reappear in new forms.
If a machine can:
learn,
adapt,
optimise,
generate novel responses,
and appear goal-directed,
should it be considered alive?
Many contemporary discussions approach this question through behaviour or intelligence. APS reframes it at a deeper organisational level.
The central question is not:
How intelligent is the system?
but:
Does the system sustain its own viability through self-maintaining organisation?
This shifts the debate away from appearance and toward the organisational conditions of life itself.
APS Box — Agency, Cognition, and Sentience: What Must Not Be Confused
Biological explanation becomes unstable when Agency , Cognition , and Sentience are treated as interchangeable. In APS, they refer to different biological or experiential attributions, and evidence for one does not by itself establish the others.
Agency is viability-oriented activity through which living organisation sustains the conditions of its persistence. It is therefore fundamental to life, but Agency alone does not establish Cognition. Biological Evaluation, Significance, and Integration may organise activity in increasingly flexible ways without necessarily crossing the cognitive threshold.
Cognition is a further attribution. Biological agency becomes cognitive when integrated biological significance modulates activity across a temporal field of viability-relevant possibilities in ways not exhausted by immediate or fixed regulation. Cognition therefore cannot be inferred merely from responsiveness, regulation, evaluation, learning-like change, or other capacities that may also admit an adequate non-cognitive explanation.
Sentience , by contrast, concerns subjective experience. Neither Agency nor Cognition, as APS defines them, is sufficient evidence that a system experiences, feels, or is aware. Cognitive organisation may therefore be investigated without assuming Sentience, just as viability-oriented Agency may be investigated without assuming Cognition.
The distinctions are consequently both conceptual and evidential. Evidence for Agency does not automatically establish Cognition, and evidence for Cognition does not automatically establish Sentience. Keeping these claims separate allows biological organisation, cognitive activity, and subjective experience to be investigated without importing properties from one into another.
Key Point. Agency does not entail Cognition, and Cognition does not entail Sentience. Each attribution requires evidence appropriate to the phenomenon being explained.
Modern AI systems can produce behaviour that strongly resembles agency.
They can:
learn from experience,
adapt to changing inputs,
optimise performance,
coordinate across large datasets,
and generate context-sensitive outputs.
To human observers, such behaviour often appears purposeful or intelligent.
APS does not deny this.
However, APS distinguishes between:
behavioural sophistication,
and viability-oriented organisation.
These are not the same thing.
A system may simulate agency behaviourally without possessing the organisational conditions required for biological agency itself.
This distinction is essential because behaviour alone does not determine whether something is alive.
APS Box — Perturbation Reveals Organisation
In APS, perturbation can be diagnostically useful because disruption may make organisational dependencies visible that are difficult to distinguish under relatively stable conditions.
A system observed only during successful functioning may not reveal which relations contribute to maintaining that functioning. Perturbation can establish contrasts through which regulation, compensation, repair, reorganisation, dependence, and failure become empirically accessible.
Under disruption, a system may degrade, remain dependent upon external support, compensate partially, or reorganise its activity in ways that maintain or re-establish conditions required for continued functioning. Comparing these responses can reveal dependencies, regulatory capacities, limits of viability, and relations involved in maintaining living organisation.
Perturbation is therefore a useful investigative strategy rather than a universally privileged diagnostic method. Not every organisational relation requires perturbation to be identified, and disruption does not by itself establish an APS explanation.
Where perturbation reveals that particular activities, constraints, or relations make a relevant difference to viability, those contrasts can provide evidence for an organisational explanation. Constraint closure, where relevant, concerns the mutual maintenance of constraints involved in sustaining living organisation rather than independence from environmental or material support.
Key Point: Perturbation can make organisational dependencies experimentally tractable by revealing how a living system responds when conditions affecting viability are altered. Whether those dependencies provide explanatory gain must still be established relative to the explanandum and relevant comparator.
APS Box — Definition, Diagnosis, and Evidence
In APS, definition, diagnosis, and evidence must not be conflated.
Definition concerns what life is :
Life is viability-oriented, constraint-closed organisation.
Diagnosis concerns how such organisation may be identified in particular systems. Relevant investigation can examine how a system maintains and re-establishes the conditions of its viability, including its responses to perturbation, vulnerability, and changing conditions. These observations do not themselves constitute the definition of life.
Evidence consists of observations that support or challenge an inference that the relevant organisation is present. Depending upon the system and question, such evidence may include endogenous repair, coordinated regulation, metabolic integration, adaptive reorganisation, and viability-oriented modulation of activity. None of these features defines life by itself.
APS therefore treats biological evidence not as a checklist of traits but as evidence bearing on claims about living organisation. A system may move without being alive, regulate without thereby satisfying the APS definition of life, or display complex behaviour without thereby demonstrating biological agency. Likewise, evidence of apparently viability-oriented activity must be interpreted in relation to the organisation of the system rather than treated as an independent definition of life.
Organised persistence poses a related but distinct explanatory problem: it denotes the continuity of living organisation through ongoing material and structural transformation. Evidence concerning repair, regulation, compensation, and reorganisation may therefore help investigate organised persistence without making organised persistence a second definition of life.
The evidential significance of any proposed indicator depends upon the claim being tested. APS terminology or conceptual architecture alone cannot establish that an observed process is diagnostic of life, biological agency, or explanatory superiority over an alternative account.
Key Point: Definition specifies the APS account of life; diagnosis concerns how the relevant organisation is identified; evidence bears on whether particular diagnostic or explanatory claims are supported. These roles must remain distinct.
APS Box — Realization Is Organisational, Not Merely Functional
In classical functionalist accounts, two systems may count as equivalent if they perform sufficiently similar causal or computational roles. APS argues that this criterion is insufficient for biological explanation because it abstracts away from the organisational conditions that make living activity possible.
In APS, realization is not understood primarily as abstract input-output equivalence. Rather, realization concerns participation within viability-oriented, constraint-closed organisation.
Biological systems are not merely collections of mechanisms performing isolated functions. They are temporally extended organisations whose processes contribute to sustaining the conditions of their own persistence. Functions therefore derive explanatory significance from their role within organised self-maintaining systems.
For this reason, materially distinct systems may realise comparable biological capacities only insofar as they participate in sufficiently similar forms of persistence-maintaining organisation. What unifies realizations is not identical material structure, nor abstract computation alone, but organisational participation within living systems.
APS therefore rejects both strict substrate essentialism and unrestricted substrate neutrality.
A system does not become biologically equivalent merely because it reproduces selected behavioural outputs or formal causal patterns. A computational simulation of metabolism, for example, does not thereby become metabolically alive, and a system that reproduces selected behavioural features associated with cognition does not necessarily instantiate biological cognition unless those activities contribute to the persistence-maintaining organisation of the system itself.
APS consequently treats realization as organisationally constrained. Biological capacities may be implemented through diverse mechanisms and structures, but those realizations remain governed by the requirements of viability, persistence, and constraint closure.
APS defines life as viability-oriented, constraint-closed organisation.
A living system:
actively maintains the conditions of its own persistence,
regenerates the constraints that sustain its organisation,
and reorganises itself under perturbation in ways that preserve viability.
Life therefore depends upon:
self-maintaining organisation,
endogenous regulation,
material persistence,
and ongoing processual continuity.
It is not fundamentally defined by:
intelligence,
learning,
computation,
complexity,
or behavioural performance.
A bacterium qualifies as living not because it is intelligent, but because its organisation continuously sustains itself against breakdown.
APS Box — Viability and Persistence
Viability and persistence are closely related in APS but distinct. Viability refers to the organisational condition under which a living system can continue to exist—the organisation that distinguishes what supports persistence from what undermines it. It defines what matters to the system.
Persistence is the ongoing achievement of that condition through time. A living system persists only insofar as it actively maintains and modulates the organisation that sustains viability under changing circumstances.
The distinction is therefore not between state and process but between condition and enactment. Viability identifies the organisation that must be sustained; persistence names the activity through which it is sustained.
Key Point. Viability is the condition that makes continued existence possible; persistence is the ongoing activity that realises it.
From an APS perspective, current AI systems do not satisfy the organisational conditions required for life.
Large language models, robotics systems, and reinforcement-learning agents depend entirely upon externally maintained infrastructures:
electrical power systems,
hardware manufacturing,
cooling systems,
server maintenance,
software engineering,
and human oversight.
Their operational conditions are externally scaffolded rather than internally sustained.
Current AI systems do not:
maintain their own metabolic basis,
regenerate their own material organisation,
repair their own physical substrate,
or preserve viability through endogenous self-maintaining processes.
Their “goals” are externally specified.
Their reward structures are externally imposed.
Their persistence does not arise from intrinsic organisational vulnerability.
For this reason, APS distinguishes:
optimisation,
from viability-oriented persistence;
behavioural adaptation,
from biological agency;
and simulated normativity,
from intrinsic normativity.
The issue is therefore not whether AI systems are sophisticated.
The issue is whether they are organisationally self-maintaining.
One of the most important APS distinctions is between behaviour and organisation.
A system may appear highly agent-like while lacking the organisational conditions required for intrinsic agency.
For APS, biological agency does not consist merely in:
responding,
optimising,
selecting,
or generating outputs.
It consists in the active maintenance of viability through organised persistence.
This means that the defining question is not:
“What behaviour does the system display?”
but:
“Does the system actively sustain the conditions of its own continued existence?”
This distinction prevents APS from:
reducing life to behaviour,
anthropomorphising machines,
or treating intelligence as equivalent to biological organisation.
APS Box — Behaviour Is Not Enough for Life
Observable features such as movement, growth, replication, or response to stimuli are often taken as evidence of life. However, these features can occur in non-living systems.
The key question is not whether a system behaves, but whether its activity is organised to sustain its own viability.
| Observable Feature | APS Interpretation |
|-------------------|-------------------|
| Movement | Is movement regulated for persistence? |
| Growth | Is growth constraint-regulated and self-maintaining? |
| Replication | Is replication embedded in self-sustaining organisation? |
| Adaptation | Does adaptation restore viability conditions? |
Key Point:
APS also distinguishes cognition from life itself.
Artificial systems may exhibit:
information processing,
adaptive coordination,
prediction,
and forms of functional cognition.
APS does not deny this possibility.
However, cognition alone is insufficient for life.
A living system must also possess:
viability-oriented organisation,
endogenous normativity,
and self-maintaining persistence across time.
This is why APS separates:
cognition,
agency,
and consciousness
as distinct dimensions of organisation rather than treating them as interchangeable concepts.
A system may:
exhibit agency without consciousness,
cognition without consciousness,
or sophisticated computation without biological agency altogether.
A common response is:
“What if AI rewrites its own code?”
From an APS perspective, self-modification alone is still insufficient.
The central question remains organisational:
Does the system sustain its own material conditions of persistence?
Does it regenerate the constraints that maintain its organisation?
Does perturbation trigger endogenous restoration directed toward viability?
Software modification, by itself, does not establish biological agency.
A system may alter its own architecture while remaining entirely dependent upon externally maintained conditions of existence.
APS therefore distinguishes:
self-modification,
from self-maintaining viability.
APS does not rule out artificial life in principle.
What matters is not biological origin but organisational structure.
An engineered or synthetic system could qualify as living if it:
regenerated its own organisational constraints,
maintained its own material persistence,
reorganised itself under perturbation,
and sustained viability through endogenous processes.
This would require:
energetic autonomy,
material self-maintenance,
organisational regeneration,
and viability-oriented regulation.
The critical issue is not whether a system was designed.
The issue is whether it becomes organisationally self-sustaining.
APS therefore allows the possibility that genuinely artificial living systems could one day exist.
However, such systems would need to cross a threshold far deeper than intelligence or behavioural sophistication alone.
Synthetic biology occupies a particularly important borderline case.
Artificially engineered cells, protocells, or hybrid biological systems may satisfy APS conditions for life if they:
maintain metabolism,
regenerate organisational constraints,
and sustain viability across time.
APS therefore evaluates synthetic systems organisationally rather than historically.
Origin does not determine life.
Organisation does.
This differs from many popular discussions in which biological and artificial categories are treated as mutually exclusive.
APS instead asks:
Does the system sustain itself as a viability-oriented process?
Many contemporary debates assume that sufficiently advanced intelligence will eventually become indistinguishable from life.
APS rejects this assumption.
The issue is therefore not whether artificial systems can become increasingly sophisticated, but whether sophistication alone explains what makes a system alive.
APS Box — What APS Changes and What It Does Not
APS changes the explanatory questions that become explicit when living organisation itself is the target of inquiry.
It proposes that Life is viability-oriented, constraint-closed organisation and examines how that organisation is enacted and maintained through biological activity. The corresponding problem of organised persistence concerns how living organisation maintains and re-establishes continuity through change.
This does not require biological explanation to begin from APS or to reconstruct other explanatory approaches within it. Mechanistic, evolutionary, dynamical, computational, informational, control-theoretic, organisational, and other approaches may provide substantive, sufficient, or preferable explanations for particular biological questions.
APS therefore changes neither the scientific status of biology nor its continuity with chemistry and physics. It proposes no special life substance, separate physical laws, non-natural causal forces, or distinct realm of living matter.
Its distinctive commitments concern how living organisation is characterised. Biological agency is viability-oriented organisational activity. Biological normativity is viability-relative asymmetry. Biological function is present-tense viability-relative organisational contribution. Purpose concerns viability-oriented organisation without requiring intention.
Agency, Process, and Scale provide complementary analytic projections of one viability-oriented, constraint-closed organisation rather than independent components, causes, dimensions, levels, or a hierarchy.
These commitments provide an explanatory framework within which biological questions can be formulated and compared. Their coherence, integration, biological specificity, or breadth does not by itself establish explanatory superiority over alternative approaches.
Where APS and another framework address the same explanatory target, their relative contribution requires target-matched assessment.
Key Point: APS makes living organisation an explicit explanatory target. It does not thereby make other forms of biological explanation subordinate to APS or establish that APS provides the preferred explanation of every biological phenomenon.
Life is not fundamentally:
computation,
optimisation,
prediction,
representation,
or intelligence.
It is organised persistence.
This reframing shifts attention away from:
behavioural imitation,
anthropomorphic appearance,
and conversational sophistication,
toward:
material organisation,
endogenous regulation,
vulnerability,
persistence,
and self-maintaining viability.
The defining boundary is therefore organisational rather than computational.
APS Box — Borderline Cases: Viruses, Sterile Organisms, Artificial Systems
Borderline cases are often taken to show that life cannot be clearly defined. In APS, they provide demanding cases against which the applicability and explanatory adequacy of a definition of life can be assessed. The relevant question is not simply whether such systems possess particular traits, but what organisation they materially realise and whether that organisation warrants classification as viability-oriented and constraint-closed.
Viruses, for example, depend extensively on host organisation for replication and continued propagation. This dependence makes them an important boundary case for APS, but dependence alone does not settle their living status. Their classification requires examination of what organisation is realised by the virus–host relation and what conclusions the available evidence supports.
Sterile organisms illustrate a different point. Reproductive capacity is not by itself necessary for living status. An organism may be unable to reproduce while continuing to maintain the viability-oriented, constraint-closed organisation characteristic of its ongoing life.
Artificial systems provide another important test. Artificial origin does not determine biological status. An engineered system should be assessed by the same organisational criteria as any other candidate system: what matters is the organisation materially realised, including whether the system itself maintains and re-establishes the constraints and conditions required for its continued functioning. External dependence is likewise not automatically disqualifying, since living systems also depend upon environmental conditions and resources.
Borderline cases therefore test how the APS definition of life applies where biological status is scientifically difficult to determine. They can clarify which organisational dependencies matter to a proposed boundary without establishing that viability orientation or constraint closure provides a uniquely APS or comparator-superior discriminator.
Key Point.
APS does not claim that artificial systems could never become alive or conscious.
Nor does it assume that biology possesses some mystical property inaccessible to engineered systems.
Instead, APS specifies the organisational conditions that would need to be satisfied before such claims become meaningful.
Whether future systems could:
sustain themselves autonomously,
enact intrinsic normativity,
or generate genuinely self-maintaining organisation
remains an open empirical and philosophical question.
APS therefore rejects both:
simplistic dismissal of artificial systems,
and premature claims that intelligence alone constitutes life.
Artificial systems may transform civilisation.
They may surpass humans in many forms of cognitive performance.
They may display extraordinary behavioural sophistication.
But from an APS perspective, complexity alone is not life.
Life begins where organisation:
actively sustains itself,
regenerates its own constraints,
and maintains viability as an intrinsic condition of persistence.
Until artificial systems cross that organisational threshold, they remain powerful artefacts rather than living agents.
The boundary of life is organisational, not computational.
APS distinguishes behavioural sophistication from viability-oriented organisation. Intelligence, learning, and adaptive behaviour do not by themselves constitute life. Living systems are defined by the organised, self-maintaining processes through which they actively sustain their own persistence across time.