Monday, March 18, 2013

Do Smart Devices Make us Smarter?

Advances in computing have removed a significant amount of pressure from the human brain. Computers provide us with quick access to nearly an unlimited source of information with websites like wikipedia, imdb, khan academy, w3school, etc, to the point where human memorization is significantly less important than in previous decades. Computers also grant us access to powerful query answering tools such as wolframalpha and google as well as calculators of various sort that out-preform TI-89s, alleviating the need for 'manual' human research or computation. Moreover, computers make it possible for all of us to access communal pools of brainpower: forums where many minds come together to solve problems; places that exemplify the old adage "two minds are better than one". These include websites like stackoverflow, quora, physicsforum, etc.

While I believe that these peripheral upgrades to human intelligence are a great improvement to our species in general, I disagree with authors like Daniel Lemire who claim that they are sufficient (or even superior) substitutes to 'actual' intelligence. In his blog, Lemire makes the claim that China's attempt to increase its population's average intelligence is an unfruitful endeavor because the pace at which intelligence can be increased genetically is grossly overshadowed by the pace at which intelligence can be enhanced using computing. I agree insofar that computers will enhance our intelligence quantitatively more and in more rapidly increasing ways compared to genetic engineering, but I disagree with the idea that we shouldn't attempt to increase our "base-line" intelligence. 

My reasoning is that increasing our biological brain's intelligence is qualitatively different from adding peripherals to our intelligence. This is actually a fairly difficult position to maintain, especially if you're familiar with the Chinese room thought experiment. In essence the thought experiment can be distilled to the following. Say you have a man who can't speak any language but English in a room with a set of perfect instructions for translating Chinese into English and English into Chinese. The room is sealed, and the only way to communicate with it is by passing notes through a slot in the door. People from outside the room may pass papers with Chinese writing into the slot, and the man inside will carry out the instructions to decipher the text, compose a response, follow instructions to translate it back into Chinese, and then pass the paper back to the outside world. (Note that we must assume such a set of instructions exists if we are to believe that Strong AI is possible, otherwise we would be forced to conclude that computers are incapable of understanding natural language.) The question then is: Can it be said that the room understands Chinese? What if the room is replaced with a Strong AI who uses those same instructions as part of its programming, does the AI understand Chinese? Is there a difference between the two?

Now picture the following thought experiment I cobbled together based on Searle's Chinese room. Say you are forced to take the final exam of a course in a subject that you've never studied before. Assume further that you're allowed to use as many smart devices during the exam as you wish, but for all other students the exam is closed-book. Then there's a good probability that you will perform better than many students in the class merely because of your access to computing devices. Can it then be said that you are more intelligent in this particular subject than the students who you out-performed? 

The point of me posing this question isn't to say that it's obviously not the case that you are more intelligent than the students whose grades you beat (because it's not obvious at all), but rather to show that the idea of intelligence is nebulous. Do we simply define the intelligence of a person as the intelligence of his or her physical brain, or should we consider intelligence as the emergent property of a system of smart devices only one of which is the biological brain?

In any case, it's my assertion that increasing biological intelligence will increase intelligence in a fundamentally different way than increasing peripheral intelligence (at least with the technology available today). 

Allow me to illustrate this with yet one more thought experiment. Take a particularly cherished children's film (a book or anything similar will work too) that you enjoyed as a child. Remember the joy you experienced when viewing the film as a child and compare it to when you viewed it as an adult. With an adult-level intelligence, you have the capacity to appreciate the film at a much deeper level then you ever could as a child no matter what contemporary smart device your child counterpart is equipped with. This definitively shows that there are certain ideas and concepts which cannot be 'translated' down to our Cartesian theater by peripheral devices the way that languages can be translated from one to another; instead they must be understood at a fundamental level within our brain.

To clarify my point a little, I am not arguing that "biological intelligence" is somehow superior to artificial intelligence. In fact, I believe that the ultimate route to expanding human intelligence is through computation rather than biological enhancements. My argument is only concerned with technology at the present. In other words, with current technology our intelligence can only be augmented in asuperficial way--for the lack of a better word--that is fundamentally different (and in many cases inferior to) biological increases in intelligence. However, note that this doesn't say anything about which type of intelligence is more useful. It's entirely plausible that a doctor equipped with a myriad of smart devices will be a better diagnostician than a doctor with only a high IQ. Furthermore, few would argue that the combination of the two is not the obvious choice if presented with the possibility. I.e., why not have both if we can?

At present, technology can make us appear to understand more than we actually do, much like the Chinese room appears to understand Chinese, but so far technology can't make the man inside the room understand Chinese; that's a job for genetic engineering, at least for the time being. 

No comments:

Post a Comment