I think that the artificial intelligence will be an important and dominant factor in human society in the near future. The popular AI programs of today (AlphaGo, DeepBlue, Watson, etc.) show great promise as building blocks for more general AI systems that could rival human intelligence. The main two problems of the current contenders in AI are high specialization and low adaptability. These systems usually excel at a single, specific task; but lack broader intelligence needed to perform outside of their task area. To me, these systems seem to correspond to specialized neurons or subregions of the brain. For that reason, I think they work well as "cells" or building blocks toward more general systems. Abstraction of these units in a similar way to ensemble methods could provide more room for generalization. Furthermore, I think a larger focus on AI programs using other AI as tools and for evaluation metrics is on the near horizon.
Despite all of this progress in AI, I still question if artificial intelligence will be able to mimic human intelligence. For me, human intelligence isn't just about logic and reasoning skills; I think other elements of the human condition such as passions, suffering, and moral values (such as altruism and sympathy) all play significant roles in abstract thought and intelligence. In particular, I think that human intelligence is so powerful based on our ability to connect disparate, seemingly unrelated experiences to the task at hand.
Because of this, I think that the Turing test offers an incomplete view of intelligence. While the AI program may be able to produce the "right" responses to sound human enough under the test conditions, there is no guarantee that the machine is processing the information in a human-like way (Chinese Room problem). Thus, I think more efforts should be made to understanding the structures of the human brain and artificial neural networks. Currently, both artificial and natural intelligence systems are largely black boxes: while they may produce desirable results, we have very little knowledge about the meaning of the connections producing the outputs. Without this understanding, I think that we will not be able to create a machine that is truly equivalent to a mind.
With this in mind, I think it's fair to say that automation and AI offer great promise as an aid to humans rather than as a replacement. We've already seen how much computers have been able to greatly increase the productivity of a single person and improve quality of living. I think AI could also offer much of the same promise. Of course, as seen with the information revolution, automation can greatly affect the role of humans in the workplace. I don't necessarily think this is a bad thing though. I feel there are great areas for growth in uniquely human areas such as care, education, and design. I don't think that we should arbitrarily throttle AI advancement or automation in order to maintain blue collar jobs; rather, we should focus more on making sure that those affected have support systems and training programs for finding work in an industry sector with more need for humans. This will require both government funding/regulation and industry guidelines for managing a changing workforce in the face of technological advancement.
Finally, I believe that the decreasing human need in the workforce due to automation pushes us to reevaluate the economic model of our country. With more and more of corporate productivity being supplied by automation, I think we need to consider some sort of Universal Basic Income in order to redistribute these productivity gains back to the populace.
No comments:
Post a Comment