# From Mimicry to True Intelligence (TI) -- A New Paradigm for Artificial General Intelligence

arXiv: https://arxiv.org/abs/2509.14474 · HTML: https://mlsanity.com/paper/2509.14474/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 13 of its 89 references that are in the map.
- First posted on arXiv 17 Sep 2025.

## Builds on (most cited first, 13 of 13)

- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [Tree of Thoughts: Deliberate Problem Solving with Large Language Models](https://mlsanity.com/paper/2305.10601/) (2023)
- [Curiosity-driven Exploration by Self-supervised Prediction](https://mlsanity.com/paper/1705.05363/) (2017)
- [Neural Turing Machines](https://mlsanity.com/paper/1410.5401/) (2014)
- [Large-Scale Study of Curiosity-Driven Learning](https://mlsanity.com/paper/1808.04355/) (2018)
- [Automatically Correcting Large Language Models: Surveying the landscape of diverse self-correction strategies](https://mlsanity.com/paper/2308.03188/) (2023)
- [ARC-AGI-2: A New Challenge for Frontier AI Reasoning Systems](https://mlsanity.com/paper/2505.11831/) (2025)
- [ARC Prize 2024: Technical Report](https://mlsanity.com/paper/2412.04604/) (2024)
- [Levels of AGI for Operationalizing Progress on the Path to AGI](https://mlsanity.com/paper/2311.02462/) (2023)
- [Hard Choices in Artificial Intelligence](https://mlsanity.com/paper/2106.11022/) (2021)
- [Metacognition is all you need? Using Introspection in Generative Agents to Improve Goal-directed Behavior](https://mlsanity.com/paper/2401.10910/) (2024)
- [Innateness, AlphaZero, and Artificial Intelligence](https://mlsanity.com/paper/1801.05667/) (2018)
- [Unsocial Intelligence: an Investigation of the Assumptions of AGI Discourse](https://mlsanity.com/paper/2401.13142/) (2024)

## Abstract (arXiv)

The debate around Artificial General Intelligence (AGI) remains open due to two fundamentally different goals: replicating human-level performance versus replicating human-like cognitive processes. We argue that performance-based definitions are inadequate, offering no roadmap for research and failing to define the qualitative nature of genuine intelligence. Four decades of work on cognitive architectures have already characterized the relevant mechanisms; our claim is that this knowledge is not being consulted by the research program now driving AGI development. What we offer is a translation of it into terms bearing on contemporary systems, with criteria for assessing them. We define True Intelligence (TI) as a system characterized by six components: five architectural pillars for which assessment criteria can be stated (embodied sensory fusion, core directives, dynamic schemata, a highly interconnected multi-expert architecture, and an orchestration layer), together with Interconnectedness, an emergent quality for which we can offer no independent measure. Two commitments distinguish the framework from the architectures on which it builds: that the topology of the schema network is itself shaped by the directives under which a system has operated, so that identical experience under different directives yields differently organized knowledge; and that a continuously active associative substrate, from which a salience mechanism promotes only a fraction to deliberate processing, is a distinct locus of cognitive work rather than the unconscious support of a task cycle. Based on the five assessable pillars we propose a five-level taxonomy of AGI, applied to four contemporary and classical systems. We hold that Level 5 is the strongest architectural condition for TI that can presently be specified, and leave open whether meeting it is sufficient.