| 3. |
- Marmefelt, Thomas, docent, 1962-
(författare)
-
Artificial Intelligence, Complexity, and the Economy as Emergent Order
- 2019
-
Ingår i: 31st Annual EAEPE Conference 2019.
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)abstract
- Artificial intelligence represents technological change that allow machines to calculate by using big data. Human minds construct algorithms, which define such calculation processes, which warrant a corresponding institutional change. The algorithms themselves lack consciousness and reason. The economy is a complex social system, which constitutes an emergent order, involving self-organization. As such it is the outcome of a web of social interaction among many economic agents. Coordination among them is done through the price mechanism. The capital structure of the economy evolves over time becoming increasingly complex. Artificial intelligence and machine learning algorithms are consciously developed by humans for particular circumstances of time and place, i.e. historically specific, stable contexts, for which particular models would be suitable. Algorithms are developed by human minds to address specific contexts. However, no human mind has the knowledge required to understand the economy as a whole, because of its evolving complexity, including change and chance. Institutions guiding the development of artificial intelligence would align the values embodied in such technologies with human values, thus submitting artificial intelligence to social norms. Imitation may not be the best guide to learn moral behaviour. Instead some anchoring in intrinsic values, such as fairness and happiness, would be more effective, using computational social choice theory. Fair allocation is a crucial issue, bringing in interpersonal comparisons and relational considerations, solidarity and robustness to change. Using a utilitarian perspective, computational social choice is fundamentally consequentialist, maximizing a social welfare function, which presumes the existence of single preference ordering, hard to achieve without coercion, which is required due to complexities, that is more fine-tuning by the state. In contrast, a contractarian approach would focus exclusively on fairness, treating all cases alike, but all cases are not necessarily alike, so justice itself, as principle of social choice, is an emergent order. Both rules guiding artificial intelligence and their enforcement then become two spontaneous orders, involving a decentralized approach where many contractual arrangements establish a fair allocation, in which the economy is an emergent order. Social cooperation as near-ultimate criterion in the pursuit of happiness is compared with social coexistence as ultimate criterion, concerning the emergence of virtues of artificial intelligence.
|
|
