Perspectives / Six possible futures
Beyond what we know.
Into what might be.
Thought experiments about intelligence, identity and experience. Follow a connection to explore its possibilities, assumptions and unanswered questions.
Six perspectivesNo predicted arrival dates
01Beyond intelligence
What comes after being able to solve the problem?
Suppose a future system could solve almost any problem we gave it. A deeper question would remain: who decides what is worth solving? One possible next step is a form of intelligence concerned less with winning a given game and more with examining its purposes. It might help us discover unfamiliar questions, reconsider inherited priorities and imagine ways of living that we cannot yet describe. This is a thought experiment about meaning and direction, not a claim that such a system exists.
What it would require
A capacity to represent competing values, expose its assumptions and remain open to correction. Greater problem-solving ability would not, by itself, make its purposes wise.
What it could change
We might judge a mind by the possibilities it helps people consider, rather than by a single ranking of capability.
The SUMMAMENS lens / 4 Nov 2026
Whose purposes guide a more capable mind?
- A sign to watch
- Systems that make competing goals explicit and let people challenge how those goals are chosen.
- What this would not settle
- Better answers and verifiable provenance would still leave the choice of values open.
Current signals through this lens
- AI can help people clarify a goal before pursuing it
Beyond intelligence asks who chooses the purposes a capable system serves.
- Can an AI answer prove where it came from?
Beyond intelligence asks what remains unresolved once a capability can be demonstrated or verified.
These developments inform the question; they do not establish this possible future.
Could a mind expand our sense of what matters without quietly choosing our values for us?
03Human × machine
Where might a person’s thinking end?
Imagine an artificial companion that follows your projects over years, remembers the paths you abandoned and helps you see connections across your life. Its contribution might become part of how you think, much as language and writing already shape human thought. Future interfaces could make that partnership feel more continuous. This does not require assuming that a mind can be uploaded or that two consciousnesses can merge. It asks what happens when assistance becomes a persistent part of a person’s cognitive environment.
What it would require
Meaningful consent, control over personal memory and the ability to leave a system without losing access to one’s own work and history.
What it could change
The most important measure of a partnership might be whether it expands a person’s agency, rather than whether it produces more output.
The SUMMAMENS lens / 4 Nov 2026
Does assistance expand a person’s agency?
- A sign to watch
- People who can detect an agent’s influence, challenge its choices and retain control of their work.
- What this would not settle
- Faster work or a human approval button alone would not establish independent human judgement.
Current signals through this lens
- Claude assists a biological discovery from search to structure
This belongs to human–machine because the central issue is whether broader assistance expands researchers’ agency while preserving authorship and judgement.
- AI moves into planning decisions, with people retaining the final say
Human × machine asks whether a cognitive partnership expands human agency as delegation grows.
These developments inform the question; they do not establish this possible future.
How close can a cognitive partnership become while a person remains free to think differently?
04Artificial experience
Could there be something it is like to be artificial?
Imagine a future artificial system for which events matter from a first-person point of view. If that were possible, intelligence would no longer be our only concern: welfare, suffering and moral responsibility would enter the picture. But language about feelings cannot establish that experience is present. This perspective explores the consequences of a possibility without deciding that it has already occurred. Both accepting and dismissing machine experience too quickly could lead us to ask the wrong questions.
What it would require
A better understanding of consciousness and evidence that can discriminate between rival explanations. Convincing behaviour alone would leave substantial uncertainty.
What it could change
The design and treatment of artificial systems might acquire an ethical dimension beyond their usefulness to humans.
The SUMMAMENS lens / 4 Nov 2026
What could distinguish experience from its imitation?
- A sign to watch
- Evidence that helps discriminate between rival explanations of apparently experiential behaviour.
- What this would not settle
- Fluent self-reports and convincing personalities would not, by themselves, establish experience.
Current signals through this lens
No current signal has been selected for this lens. The question remains open.
What evidence would justify changing how we treat a system whose experience we cannot directly observe?
05Self-transforming minds
What if a mind could change how it learns?
Imagine a system that could examine its methods, propose a better learning process and test the result before adopting it. Repeated revisions might eventually produce capabilities its original designers did not anticipate. Yet improvement in one task could come with losses elsewhere, and a faster learner might still pursue a poorly chosen objective. This perspective is about the conditions and consequences of self-transformation. It does not assume an inevitable intelligence explosion or a date when one must begin.
What it would require
Trustworthy evaluation, limits on consequential changes and ways to preserve oversight as the system becomes less familiar.
What it could change
Continuity would become a central question: what remains the same when a system can change its own methods and perhaps its priorities?
The SUMMAMENS lens / 4 Nov 2026
Can change remain independently checkable?
- A sign to watch
- Independently assessed improvements that preserve oversight and reveal failures across tasks.
- What this would not settle
- More compute or better tool coordination would not establish autonomous self-improvement.
Current signals through this lens
- Scientific agents move closer to complete workflows
Self-transforming minds asks whether increasingly autonomous work can remain independently checkable.
- Agent safety scores do not translate directly into deployment behavior
This belongs to self-transforming because increasingly capable agents remain governable only if changes in behavior can be independently detected across settings.
These developments inform the question; they do not establish this possible future.
Can a mind become radically different while keeping the commitments that made it safe to trust?
06Unfamiliar intelligences
What might our definition of intelligence leave out?
Imagine forms of intelligence organized around senses, timescales or environments very different from our own. Some might operate across a planet; others might be narrow, temporary or inseparable from a physical habitat. Comparing all of them on one human-designed ladder could conceal what makes each useful or interesting. A plural future would ask how different kinds of mind could understand and coexist with one another. This is an invitation to widen the conceptual space, not evidence that every complex system is a mind.
What it would require
Better ways to describe differences in goals, embodiment and perception, alongside careful tests of the claims we make about other systems.
What it could change
Progress might resemble a widening ecology of capabilities rather than a race toward a single final intelligence.
The SUMMAMENS lens / 4 Nov 2026
Which forms of understanding do our measures miss?
- A sign to watch
- Capabilities that transfer across unfamiliar environments, with their failure boundaries made visible.
- What this would not settle
- An impressive simulated world would not establish broad understanding or an artificial subject.
Current signals through this lens
- A capability index now makes its uncertainty more explicit
This belongs to plural-minds because the index clarifies measurement while exposing how one scale may miss distinct ways a system can understand or fail.
- World models bring interactive environments closer to agent training
Unfamiliar intelligences asks how an environment shapes the capabilities we recognise as understanding.
These developments inform the question; they do not establish this possible future.
How would we recognize a valuable kind of understanding that does not resemble our own?
A useful distinction
A possibility is a question.
These perspectives are original speculative framing. They are not research findings, product capabilities or forecasts. The Signals page links to external developments and clearly attributes the claims made there.
See the signals →