preprint · arXiv (Cornell University) · 2023
DeepThought: An Architecture for Autonomous Self-motivated Systems
See where this sits in the topic map →Summary AI-generated
- TL;DR
- As large language models (LLMs) engage in increasingly credible dialogues, discussions have emerged around whether they might exhibit intrinsic motivations, agency, or consciousness.
- Problem
- Not specified in the abstract.
- Method
- The authors argue that the internal architecture and finite, volatile state of current LLMs cannot support intrinsic motivations, agency, or consciousness on their own.
- Results
- Not specified in the abstract.
- Contributions
- Not specified in the abstract.
- Limitations
- Not specified in the abstract.
- Takeaways
- By combining insights from complementary learning systems, global neuronal workspace, and attention schema theories, the paper proposes an architecture that integrates LLMs and deep learning systems to create cognitive language agents capable of agency, self-motivation, and features of meta-cognition.
- Applications
- Not specified in the abstract.
- Topics
- Large Language Models, Cognitive Architectures, Artificial Agency, Deep Learning
- For industry
- Not specified in the abstract.
- Why it matters
- Not specified in the abstract.
Abstract
The ability of large language models (LLMs) to engage in credible dialogues with humans, taking into account the training data and the context of the conversation, has raised discussions about their ability to exhibit intrinsic motivations, agency, or even some degree of consciousness. We argue that the internal architecture of LLMs and their finite and volatile state cannot support any of these properties. By combining insights from complementary learning systems, global neuronal workspace, and attention schema theories, we propose to integrate LLMs and other deep learning systems into an architecture for cognitive language agents able to exhibit properties akin to agency, self-motivation, even some features of meta-cognition.