Smith’s AI & Economics
Idealizations of Uncertainty, and Lessons from Artificial Intelligence Robert Elliott Smith
Economics special issue on Radical Uncertainty and Its Implications for Economics.
Economic models typically assume that human agents gather and process information about the alternative choices in any given situation.
This idea of probabilistic reasoning as the core extension to rationale under uncertainty is also reflected in the most recent developments in artificial intelligence (AI).
It is important for economists to understand the unsolved problems of AI in modelling human decision-making under uncertainty, since it is all too easy to assume that probabilistic models in AI are near to modelling human decision-making behaviour effectively.
5 Alternative Representations of Uncertainty
(Alternatives to classical logic and probability.)
Lane and Mayfield (2005) divide human uncertainty into three types: truth uncertainty, semantic uncertainty and ontological uncertainty … : uncertainty about the objects that exist in the universe of concern.
…. it is clear that humans cope robustly with ontological uncertainty on a daily basis …. [yet] probability theory does not treat ontological uncertainty.
Ontological uncertainty, and its ubiquity, preclude logical or probabilistic ideals that suggest there is a way that humans should think in reality, in that there is no set of atoms for an ideal logic that can be pre-described, and no optimal solution in probabilistic expectation over a pre-described event space. Models of human decision-making that have started with ideals of logic or probability theory (and problem domains derived from those ideals) start explicitly by eliminating the possibility of ontological certainty.
Consider conviction narrative theory (CNT) as a description of how humans cope with decision-making uncertainty (Tuckett et al., 2014)(Tuckett, 2011). CNT asserts that to cope with decision-making uncertainty, human actors construct narrative representations of what will happen in the future, as if the objects and relationships involved were certain. To continue to reason and act within these representations of the future (for instance, to continue to hold a financial asset), despite decision-making uncertainty, the human actor emotionally invests in the narrative, particularly to balance approach and avoidance emotions, such that action can be sustained. In this sense, emotion plays a role the human decision-making process by providing a foundation upon which conviction can lead to action, despite uncertainty. Tuckett’s interviews with hedge fund managers across the recent financial crisis show consistency with CNT (Tuckett, 2011).
CNT results also are consistent with the fact that human decision-making takes place in a social context. Examinations of conviction narratives in financial news (which are drawn from a large social context) are providing powerful causal results about the macro economy (Tuckett et al., 2014). These results are consistent with CNT combined with observations by Bentley, O’Brien, and Ormerod (Bentley et al., 2011), who note that in making economic decisions, humans can socially rely on experts, or on a peer group. Thus, social sharing of ideas is an important aspect of understanding human decision-making.
… Thus, an important element of re-examining the idea of human decision-making in the face of an ontologically uncertain world may be the mechanisms of innovation of human ideas in both the psychological and social contexts.
AI has shown that models of human reasoning that are based on the idealisations of mathematics or logic do not embody the robust decision-making in the face of ontological uncertainty observed in humans.
… We must consider how humans do reason under real-world uncertainty, where such idealisations do not apply. This suggests the reinclusion of observations from sciences that have focused in vivo human behaviour (psychology, sociology, anthropology, etc.). By focusing on this non-idealised decision-making, we may be able to have more realistic models of economic actors, as well as substantial advances in AI.
This is clearly of great importance, not just to economists. But as a mathematician with a logical bent I must question some of the wording. My online dictionary has two definitions of ideal:
- satisfying one’s conception of what is perfect; most suitable.
- existing only in the imagination; desirable or perfect but not likely to become a reality.
I accept that what is referred to as ‘idealised decision-making’ is an ideal for some, but my reading of the paper is that it supports my view that in fact such decision-making is abominable, and that we need to educate economists, politicians and – ideally – the public to this. I conjecture that there is an idealisation such that we can draw on more appropriate logics and mathematics, to good effect for some of our more intractable problems.