DCS-T-709-AIES-2024 Main
Link to Lecture Notes

SUMMARY OF LECTURE NOTES BY K. R. THÓRISSON

Learning & Autonomy

Empirical Learning	When information comes from measurements in the physical world it is “empirical evidence”. 'Empirical learning' is thus learning based on empirical data.
Experience-Based Learning	Learning is the acquisition of knowledge for particular purposes. When this acquisition happens via interaction with an environment it is experience-based.
Causation	Knowledge of causation is a fundamental concept in building reliable information models (knowledge) from experience.
Explanation Depends On Causation	It is impossible to explain anything in any useful way without referring to general causal relations.
Why Explanation Matters	If a controller does something we don't want it to repeat, it needs to be able to explain why it did what it did. If it can't, it means it - and we - can never find out why it did that, whether it was caused by a design flaw, and under what conditions it might do it again.
Trustworthiness	The more reliably it does its job (and nothing else) and does it well, the more trustworthy it is. The more trustworthy it is, the better it is suited to be designed to follow ethical rules and regulations.
Achieving Trustworthiness	Requires reliability, and predictability at multiple levels of operation. For autonomous machines, this requires them to know reliable causal relations.
Cognitive Autonomy	The ability of an agent to act and think independently. Implies that the machine “does it alone”. Refers to the mental (control-) independence of agents - the more independent they are (of their designers, of outside aid, etc.) the more autonomous they are. Systems without it could hardly be considered to have general intelligence.
Moral Autonomy	Cannot be assumed for any autonomous machine if that machine does not understand the side effects of its own actions.
Conclusion	Moral autonomy in machines must rely on understanding, which relies on reasoning and knowledge of cause-effect relations, which relies on autonomous learning of complex phenomena.