Page 189 - Artificial Intelligence for the Internet of Everything

Would IOET Make Economics More Behavioral? 175

what is known as the socialist calculation debate (Boettke, 2000). Ludwig
von Mises (1881–1973) and Friedrich Hayek (1899–1992) led the charge
against socialism; this debate, while alien to most computer scientists and
AI researchers, can be considered to be the longest and the largest-scale
4
debate on the possibility and desirability (consequences) of AI. Alongside
this debate, Hayek published his pathbreaking book, The Road to Serfdom
(Hayek, 1944), which is probably the earliest penetrating and thought-
provoking warning against unlimited exposure to AI. Notice that Hayek
did not object to the use of machines or chess games with robot players;
he raised objections to the use of AI to institutionalize the market
mechanism.
So, is this time different? Even though ICT in the last century was insuf-
ficient to make the Walrasian auctioneer “smart enough” to support a cen-
tralized unmanned market, the current ICT revolution is in a better position
to exhibit that function. Therefore what concerns us this time is not the
usual job-loss alarm, which does not make this time different, but rather
the market-annihilation alarm, which does. Insofar as Hayek’s warning is
valid, Helbing et al. (2017) can then be read as a refreshing message of Hayek
(1944): instead of empowering the state over individuals, technology can
now empower the platform over individuals. In any case, we could be under
the shadow of Thomas Hobbes’ (1588–1679) Leviathan (Hobbes, 1977).

10.3 HOMO ECONOMICUS VS. HOMO SAPIENS

Economics deals with the causes and consequences of human decision
(choice) making in resource allocation subject to various physical constraints
and desires. Mainstream economics has long established a standard represen-
tation for the rationality of this decision-making agent, which is normally
known as homo economicus or the von Neumann-Morgenstern expected-utility
maximizing agent (Von Neumann & Morgenstern, 1944). These agents can
learn from experiences and hence can cope with the future without making
systematic errors. They can also solve various trade-offs in their lives: con-
sumption and saving, portfolio returns and risk, education and employment,
work and leisure, diet and health, residential location and commute flexibil-
ity, numbers of kids, etc. Apart from the accidents, incidents and natural
disasters that are beyond their control, their rationality implies that they have

4
It is only very recently that the computational-theoretic nature of the debate was realized when this
issue was visited by computer scientists (Cockshott & Cottrell, 1997; Cottrell & Cockshott, 1993).