Whatever the origins of our DNA, it seems that our internal clock is geared towards a certain rate of development. As children, it takes our human brains a given amount of time to understand certain concepts, and even once we are adults, we tend to take time to adapt to change, all the more so if that change is major. Until recently, despite what seemed to be a very rapid change in the world of technology, overall our day-to-day life still had a manageable pace. Admittedly, there are still far too many people who are not comfortable with sending an email, let alone using an instant messenger or more advanced computer-based tools. But these individuals still function in our present communities, and still are able to stay abreast of important news and updates about their health and welfare.
I consider myself well versed in various technologies and I openly welcome even rapid advancements in digital tools, no matter how sweeping. But yesterday, I viewed a TED talk which truly gave me pause. And based on some recent comments even on high tech sites, I believe that many people, if not most of the world, is ready for the upcoming transformation in our lives.
For those who are not acquainted with TED talks, they are a brilliant series of lectures, freely available on the Internet. This actually is a beautiful example of the Internet at its best. Top-of-their-field individuals speak for approximately half an hour on any topic they wish. To be clear, these speakers are very carefully selected and undergo a rigorous vetting process before being given stage. that is one of the reasons that the quality of every lecture is so high.
I would ask you to stop reading at this point and take the time to watch the specific TED talk I am referring to. this particular lecture is delivered by Jeremy Howard, who is a top name in data analysis and artificial intelligence. For approximately 20 minutes, Mr. Howard speaks of a relatively new system for computer learning. This new system is called “deep learning” and truly demonstrates one of the key features of a successful artificially intelligent system – once the AI system has learned the basics, it can exponentially increase its understanding.
I should point out that such AI systems, that can learn on their own and thus surpass certain human mental capabilities, are not designed in the same way as standard software. No programmer has written lines of software code that tell Deep Learning systems how to respond when the system “sees” a given image. In software I write, I occasionally ask a user to upload an image, which I then manipulate with other software and save on the server. This very straightforward kind of code, which can easily be traced through its various steps, simply does not exist within advanced AI systems.
One could easily ask, if you can’t trace the steps of the decision process being made by AI systems, how do you fix bugs? Well, it may not be so easy. If there is a fundamental problem in the way internal algorithms are created by the AI system, you may be able to make a difference. But if there is a problem with the decision that the AI system made based on these algorithms, generally speaking, the only solution is to teach the AI system more.
This should sound familiar. This is the way in which we teach humans. If a professor is teaching a class about calculus, and a particular student fails an exam, no one considers performing neurosurgery in order to fundamentally change the learning center in the student’s brain. Instead, you may offer the student additional help or a tutor. But ultimately, all you have to offer is more teaching. Advanced AI systems like Deep Learning have reached this stage as well.
During the TED talk, a few examples are presented of how Deep Learning can surpass humans in certain decision-making situations. These are not trivial examples. It is evidently clear that such an AI system will soon replace humans in various positions, even ones considered key to human welfare. Admittedly, talk about self driving cars has increased dramatically in the recent year. And over the course of the next decade, self driving cars will become much more common and will eventually replace present-day automobiles. I have said before that my children’s children will most likely never learn how to drive. At the same time, they will benefit from safer driving than anyone experiences today.
I would go so far as to say that you cannot watch this TED talk and not wonder how society will effectively respond. As I noted above, human beings can only process change at a certain rate. But technology has reached a critical point at which it is no longer ready to wait for humans to catch up. I truly believe that it is no longer an exaggeration to expect major upheaval in the human employment situation, over the next 20 years. Even today, unemployment rates amongst young people, freshly out of university, are frighteningly high. If AI systems do succeed in all of the areas that they can affect, I personally am not sure what job opportunities will still be available for most people.
I read another article yesterday, posted on the Singularity website. I have mentioned this website in the past, and it is a bringing together of top minds and opinions in the world of technology. I was actually very surprised by this article because, quite frankly, I really think it is totally wrong. Allow me the following quote from this article:
“The most influential people will be those who can wade through large amounts of information to spot problems, trends and opportunities, see the immediate task as well as the big picture, come up with original ideas in an increasingly noisy world, and then build and manage teams, networks and ecosystems to address problems and opportunities in a way that creates value and fairness for all who engage.”
Initially, this one statement sounds utopian. But it is inconsistent with what we are already seeing. Human beings will not be the ones to wade through information to spot trends. Even today, the most successful human beings are those who can pass over this responsibility to computerized analytics. I agree that humans are still driving the discovery process. But the legwork is already being done by computers. A system like deep learning will eliminate even this degree of human involvement. After the initial learning period, humans will stand by and watch what the AI systems report as being significant trends. It will not even be possible to double check the work of the AI system, simply do to the mass of data that formed the basis of the AI system’s conclusions.
I truly do not mean to sound overly cynical, but I have rarely seen an example of a Fortune 500 company based on the principle of “fairness for all who engage”. Fairness is a marvelous human quality. It is one I fundamentally believe in, and it is one of the reasons I hold to my socialist roots. But few business plans ever refer to fairness as part of their killer algorithm.
Another quote from the same article also, in my opinion, misrepresents the significance of a human quality. The author of the article states:
“we have launched a global strategy to help children around the world learn empathy and changemaking in school”
Once again, this is nothing less than a wonderful utopian dream, one that I wish I could share. I personally believe that is very difficult to teach empathy. I think that years of exposure to an empathetic environment, can bring out the empathetic side of a person. But there will naturally be some people who are more empathetic and some who are less. If our only distinguishing characteristic from computers is this difficult-to-quantify characteristic of empathy, then I think we have already handed the keys to society over to AI.
I have absolutely no magic answers for increasing participation of younger people in the job market. Based on what I have seen in this TED talk, human beings will soon be struggling to find purpose and employment in an AI driven world. The accompanying social change, i.e. more unemployed youth in the midst of an aging yet not-dying older population, will fundamentally change the way we live. How our society will look in 20 years, 50 years and beyond, is anyone’s guess. I would be very hesitant to trust any “expert’s” predictions about how this will all play out.
The purpose of an earthquake warning system is not to stop the earthquake. We don’t have the technology for that [as of yet]. The purpose of such a warning system is to allow people to leave the area or to take the necessary precautions to survive the earthquake.
Society is facing the equivalent of an earthquake that will fundamentally change its face. All anyone can do is warn the public to be ready, and to take whatever steps possible to best “weather the storm”. The difference is that an earthquake leaves only destruction in its path. AI will lead to a better world in many ways. The question is how to get humans to join along for the ride.
Thanks for listening