The Next Stage of Evolution: research notebook

How to read this notebook

This notebook accompanies “The Next Stage of Evolution”. It is a bounded starting point, not a systematic literature review or a survey of all AI capabilities as of publication. Its two research anchors support narrow claims; the essay’s longer trajectory is a thought experiment.

Established / evidenced: AlphaFold protein prediction

Claim: A human-built computational system can contribute substantially to a specific scientific task. Jumper and colleagues’ 2021 paper reports AlphaFold’s performance in protein-structure prediction, including the CASP14 assessment (the primary paper).

Boundary: The paper concerns structure prediction. It does not establish subjective experience, open-ended scientific agency, self-maintaining industry, or an independent artificial civilisation (Jumper et al.).

Cited in: The Next Stage of Evolution → A small, firm starting point

Emerging / plausible: AI consciousness framework

Research question: How might we assess the possibility of artificial consciousness? Butlin and colleagues derive provisional indicator properties from several theories, while acknowledging uncertainty about both the theories and their application to AI (the 2023 report).

Boundary: The authors’ assessment of the systems discussed in 2023 is a dated assessment, not a claim about every later system. Satisfying indicators would not constitute a definitive test, and behavioural fluency alone is not treated as sufficient evidence (Butlin et al.).

Cited in: The Next Stage of Evolution → Capability is not consciousness

Interpretation in the essay

The essay treats capability, consciousness, autonomy, and physical self-sufficiency as separate questions. It proposes that a contribution to scientific work makes further questions worth investigating, without making the answers to those questions inevitable. This is the essay’s reasoning, not a finding attributed to either source.

Speculative / hypothetical

• Systems participating reliably in the creation of increasingly capable successors.
• A self-sustaining artificial lineage with an independent developmental trajectory.
• Conscious experience in future artificial systems.
• An artificial civilisation maintaining itself beyond Earth or travelling to other stars.

These scenarios are not established by the sources below. They require additional assumptions and investigation. No probability, timetable, or inevitable outcome is assigned to them.

What would strengthen or weaken the exploration?

Useful follow-up evidence would include reproducible demonstrations of broader scientific work, independently evaluated contributions to system development, clearer accounts of consciousness indicators, and concrete analysis of physical maintenance and resource requirements. Failures, bottlenecks, dependence on human correction, and competing explanations would be just as important to record.

Future revisions should add that evidence with its limitations intact, rather than assemble only material that supports an attractive story. The source annotations below state what each source contributes and what it does not.

Source material

1. Highly accurate protein structure prediction with AlphaFold

   Peer-reviewed paper · John Jumper et al. · Nature · 2021

   A concrete anchor for the essay's narrow starting claim: computational systems can make substantial contributions to a difficult scientific task. The paper reports protein-structure prediction results, methods and evaluation.

   Limit: Does not demonstrate consciousness, general scientific autonomy, recursive self-improvement, or an independently sustained artificial civilisation.

2. Consciousness in Artificial Intelligence: Insights from the Science of Consciousness

   Research report / preprint · Patrick Butlin, Robert Long et al. · 2023

   A theory-informed framework for asking about AI consciousness without relying solely on human-like behaviour. It derives provisional indicator properties from multiple scientific theories.

   Limit: A provisional framework, not a conclusive test. Its assessment of the systems discussed is dated to 2023; it does not establish the status of later systems or predict an artificial civilisation.