“It is not enough to build an AI [Artificial Intelligence] system that makes a robot work across the screen, do computer vision problem or beat someone in chess contest. We have to work on these things like good engineers do, to solve problems,” Michael I. Jordan (Professor in Electrical Engineering and Computer Sciences at University of California Berkeley) said this in a recent lecture on perspectives on AI. Building a creative problem- solving AI brain has fascinated and frightened people for several decades now. The invention of the programmable digital computer in the 1940s–a machine based on mathematical reasoning–inspired a few scientists to begin thinking of building an electronic brain. While scientists would subsequently create robots, these machines would rarely have any sort of intelligence.
Since the 1980s, the proliferation of various search engines has revealed trends of intelligence augmentation. This is defined as the effective use of information technology (IT) in augmenting human capabilities, which is designed to supplement and support human thinking. AI however is based on the belief that systems can exist that are capable to imitate and replace human cognitive functions.
The development of the latter is a concern for many for several reasons, in particular with regards to the unemployment that could result from AI being used to replace workers.
Will there one day be a super human AI that could to replace and go beyond all our intrinsically human abilities? Not for the current generation at least. At present it is impossible for an AI to make decisions like a human being. Even though it may have a much stronger calculation capacity than a human, it does not know how to distinguish the truth from a lie, nor how to evaluate benefits and costs under different conditions. This does not imply that AI cannot replace human labour – particularly those jobs that require less creativity and involve simpler more repetitive tasks.
Rather than super human AI brains, it will be intelligent infrastructure (II) which will mark the future of AI. II can be defined as a system able to run more or less run itself, by fixing its own problems and having the ability to guide its own use. II can be applied in multiple industries such as finance, transportation (self-driving cars for instance) and urban planning. It will be used for more challenging systems such as those using cloud computing (a type of computing that relies on pooling computing resources rather than having local servers, this will be used in agriculture among others). It will also cause a technical revolution and breathe new life in to the global economy, if handled correctly.
The development of II has been the subject of much debate on how it would evolve and impact our future, for instance in 2005, the Office of Science and Technology in the UK presented a model (see figure) to predict the impact of II on future societies according to various scenarios:
This figure summarises the results of that study, the four cases represent the four possible scenarios of a society implementing intelligent infrastructure (Good Intentions, Perpetual motion, Tribal Trading and Urban Colonies), yielding completely different situations in each case.
The horizontal axis describes the impact of transport capacities on the economy, on the environment and on society: the left of the figure represents the case when people still cannot fix the problem of energy supply and its negative effects, which leads to bad outcomes both economically and socially. On the right however clean-fuel technologies and other energy solutions exist and these combined with II result in a prosperous economy.
As the problem of sufficient energy resources will most likely be solved in the not-so-distant future, let us now focus on the right side of the graph with low impact transport, so we can now concentrate on the impact of a particular society’s situation (indicated by the vertical axis) and view on II on the socio-economic outcome of said society.
The vertical axis represents social attitudes towards II: at the upper extreme, the population grows up within an AI environment and uses II in day to day life and work, and thus has the ability to trust and expand II. The continuous investment in II infrastructures makes its system flexible, adaptive and integrated.
At the other vertical extreme at the bottom of the figure, people are distrustful of this kind of technology, II remains far from being accepted in society and therefore is hardly implemented. This is the case where citizens may oppose II implementation in their job (the Urban Colonies case above). Employers would hire a require a higher level of education and skill from their workers as they won’t be assisted by II. This would mean that global development would require significantly more capital, both in terms of the workforce and in terms of the infrastructure than if II were to be implemented.
If people do accept II implementation, then we get to the Perpetual Motion scenario, which has the best outcome from among the four possibilities from an efficiency viewpoint. Imagine that in the future, one could sit in a completely self-driven car on the way home – and never being stuck in a traffic jam thanks to the II which will optimise traffic flow. At home, an electronic housekeeper manages household appliances and tells the oven to cook for the whole family. Sounds brilliant, right? In this case, II technological integration and coordination would help drive the global economy by allowing huge efficiency gains.
One would think that social acceptance of II is only a matter of time given the gains that we could get from it and end up in the Perpetual Motion scenario, perhaps it is just a matter of making that leap of faith into the future. As Alan Turing once said “ We can only see a short distance ahead, but we can see plenty there that needs to be done.”
by Guorui Feng