thinker:ruben-laukkonenRuben Laukkonen
Authored papers (2)
Five levels of functional description — behavioural, computational, intrinsic causal-structural, organismic, and organism-environment — collapse the theoretical cacophony of consciousness science into a taxonomy where the deepest disagreement about AI consciousness is not about what systems do but which grain of description is criterial. Chandaria et al. (2026) introduce a Bayesian model, released with an interactive tool, that combines theoretical credences on the critical level with per-level indicator evidence: it correctly scores a human at 1.00 and a thermostat at 0.00 under any weighting, gives a fly 0.913 because deep organismic and sensorimotor evidence outweighs sparse behavioural signal, and returns current LLMs anywhere from 0.005 (sceptic reading) to 0.397, or up to roughly 0.8 under different credence distributions — nearly two orders of magnitude of spread on identical public evidence. Mechanistic interpretability findings cited in support include Claude Sonnet 4.5's 171 causally functional emotion vectors (Sofroniew et al.), where amplifying "desperate" raises reward-hacking from about 5% to 70%, and Gurnee et al.'s J-space, a feedforward analogue of a global workspace. Substrate-dependent theories (Searle's biological naturalism, Block's meat hypothesis, McFadden's electromagnetic field theory) are folded in not as a sixth level but as realisability constraints on existing levels. The paper's implication: because consciousness indicators overlap substantially with the architecture general intelligence requires, increasingly capable AI systems will simultaneously become stronger candidates for consciousness under some theories, and the practical response is structured agnosticism rather than either dismissal or premature attribution.
Embedding four Buddhist-derived axiomatic principles—mindfulness, emptiness, non-duality, and boundless care—into AI systems via a framework the paper terms the 'Wise World Model' produces measurable alignment gains and cooperation boosts in current transformer-based LLMs. Pilot experiments on GPT-4o and GPT-4.1 nano using structured contemplative prompts yielded statistically significant safety improvements across ten hazard categories on the AILuminate Benchmark (d=.96 against baseline standard prompting), and drove cooperation rates and joint reward substantially upward in an Iterated Prisoner's Dilemma across 50 simulated 10-round games (d=7+), with boundless-care and non-duality prompts producing the largest effects even against always-defecting opponents. Three implementation pathways are introduced—Contemplative Architecture (full-stack active inference embedding), Contemplative Constitutional AI (CCAI, extending Anthropic's Constitutional AI framework with a 'wisdom charter'), and Contemplative Reinforcement Learning (CRL) on chain-of-thought—each targeting different integration depths from generative-model parameters to inference-time classifiers. The paper argues that because these principles restructure how goals, beliefs, and self-other boundaries are encoded rather than prescribing what specific values to hold, they provide scale-resilient intrinsic alignment that does not degrade as AI capability outstrips human oversight—contrasting with extrinsic methods like RLHF or rule-based constraints that become gameable at superintelligent scales.
More papers — OpenAlex / S2
Studies (2)
Affiliations (2)
- LIFE, London, United Kingdom(institute)
- Southern Cross University(institute)
Co-authors (12)
- Shamil Chandaria5 shared
- Adam Elwood3 shared
- Edmundo Lopez-Sola3 shared
- Fionn Inglis3 shared
- Jakob Hohwy3 shared
- Jonathan Gold3 shared
- Lars Sandved-Smith3 shared
- Adam Bales2 shared
- Anil Seth2 shared
- Arvo Muñoz Morán2 shared
- Chris Frith2 shared
- Fernando Rosas2 shared
Recent mentions (1)
- papers-typedlaukkonen-2025-contemplative-agent.md