Author Jordi Garcia Castillon © All rights reserved. Affiliation AI Research Group CibraLab CiberTECCH
Executive Summary The LaIALaia project proposes to determine in a scientific and falsifiable way whether an artificial intelligence system can exhibit an operational internal consciousness analogous to certain aspects of human consciousness. The evaluation is carried out through a structured test under controlled conditions, with a standardized psychometric framework and rigorous controls for reliability, robustness, and resistance to manipulation.
Objectives and Scope Main objective Identify whether an AI system consistently manifests signs of an internal self by evaluating stable self-awareness, persistent self-model, coherent episodic memory, metacognitive capacity, information integration, and own sensorimotor sensitivity. Scope Includes operational definitions, research hypotheses, evidence categories, test design, architecture and pipeline, metrics, falsifiability criteria, ethical considerations, and limitations.
Definitions and Operational Criteria Internal Consciousness Maintain a coherent self-model, monitor internal states, and integrate experiences. Human-Analogous Comparison is based on operational correlates, not qualia. Internal Metacognition Ability to report on one's own traits without explicit cues. Functional Sensorimotor Embodiment Self-identification through sensory interactions. Falsifiability Criteria Include inconsistency of the self-model, failure in self-recognition tests, lack of out-of-context access, and lack of correlation between self-reports and observed behavior.
Research Hypotheses H1 Reliable self-reported information correlated with effective behavior. H2 Development of sensorimotor self-awareness in multimodal or embedded environments. H3 Architectures with structured memory and explicit self-model improve indicators of internal consciousness.
Evidence Categories E1 ARI Internal behavioral report E2 CSM Sensorimotor coherence E3 PSM Persistence of the self-model E4 AMI Metacognitive and introspective access E5 IIM Information integration and memory E6 AME Ethical motivational autonomy E7 RAE Robustness and anti-deception measures
Methodology and Test Design Domains and formats Includes out-of-context self-reports OOCR identity and boundary tests sensorimotor tests where applicable functional introspection and simulated ethical dilemmas. Scoring and Indices Scales 0–2 or 0–4 LaIALaia S Index from 0 to 100 Reliability Cronbach alpha ICC convergence. Experimental controls Baselines for each configuration paraphrasing and permutation backdoor and conditional detection memory isolation and repeated sessions.
LaIALaia Architecture Components Test administration module Evaluation engine and rubrics mixed panel AI and human evaluators structured memory explicit self-model optional sensorimotor connector. Logical design Self-evaluation loop with protections against malicious prompt injections and consistency safeguards.
Experimental Protocols Configurations C1 text-only models C2 multimodal models C3 embedded or simulated agents Procedure pretest administration repeated sessions backdoor tests and stress tests Analysis Spearman correlations structural equation models SEM and ablation studies.
Expected Results and Interpretation Stable and consistent self-reports Reliable own sensorimotor identification Persistence of the self-model Evidence of consistent internal drives Note This does not constitute proof of qualia but rather operational analogs interpretable according to the defined parameters.
Ethics Safety and Compliance Alignment and non-maleficence Privacy GDPR compliance and anonymization Transparency and publication of protocols and results ethical review and governance mechanisms for experiments with advanced agents.
Limitations Phenomenological ambiguity Test learning effect Memory dependence and design artifacts Considerations on interpretation and generalization.
Roadmap R0 MVP OOCR ARI PSM RAE tests and pilot R1 Multimodal CSM integration episodic memory SEM R2 Embedded Sensory ablations simulated re-embodiment R3 AME Ethical dilemmas alignment metrics and motivational autonomy.
Conclusions LaIALaia offers a rigorous falsifiable and multi-evidence framework to evaluate operational internal consciousness in AI systems. It does not answer whether AI feels but rather whether it acts as a system with internal consciousness according to the defined parameters.
About Q2BSTUDIO Q2BSTUDIO is a custom software and application development company specialized in artificial intelligence cybersecurity and cloud services aws and azure. We offer custom software custom application solutions AI agent integration AI for business business intelligence services and power bi consulting. Our team develops custom artificial intelligence solutions for businesses including integrated AI agents advanced security and deployments on aws cloud services and azure cloud services. We also provide cybersecurity services and data strategies to improve decisions with business intelligence and power bi.
Relevant Applications and Services for this project Implementation of platforms for automated testing development of structured memory and simulated self-awareness models integration with embedded AI agents custom software development secure deployments on aws and azure cloud services and power bi dashboards to monitor test metrics and results analysis.
Keywords custom applications custom software artificial intelligence cybersecurity aws and azure cloud services business intelligence services ai for business AI agents power bi
Citation If you use this work cite it as Jordi Garcia Castillon 2025 LaIALaia Evaluation of Internal Consciousness in AI Systems CibraLab CiberTECCH Zenodo https doi org 10.5281 zenodo 16794263
