Who, What, How, Why, At What Cost and at What Risk (Part II)
(Is it Deja Vu all Over Again – Part VI)
SUMMARY
We now start to consider questions raised in the previous post by asking “What’s it all about Alfie?”, a flashback to a well-known movie title that seems to say it all in the present circumstances. If you have questions, we will be seeking the answers in this and subsequent posts so that we may better understand just what is happening. We’ll consider all of the related issues mentioned in our heading, who, what, how, why, at what cost and at what risk? We’ll formalize our list of questions here, and begin to answer them in this post.
Introduction
We have now arrived, after a brief historical survey, at the dawn of the present day. Here we find ourselves being heralded almost daily with news of the latest eye-opening developments in AI. And, if not yet reality, at least we’ll hear the latest speculation on what AI is going to do and what the impact will be on every aspect of what might possibly become our future existence.
If we believe all that we are told, we earthly mortals may simply become superfluous at the dawn of some new day lurking dimly in our future. And what hope can there be, as we ordinary humans come to rely more and more on the intelligent machine of one kind or another, to do our thinking and perhaps even our work for us? Will it be nirvana finally achieved, or perhaps only a dark and dismal existence from which we will seek reprieve? Unless we know and understand more about what we are actually dealing with, these questions might be difficult to answer.
In light of the ever present hype, once again commonly refered to as “AI”, or “artificial intelligence”, we propose to learn more about the players, the technology itself, and the aspirations and questions associated with this latest incarnation of the monster that just might eat human beings!
Before we dive into the AI pool too deeply, it is always a good idea to review what we are looking for and what questions seem to be most relevant to understanding and evaluating what we are actually dealing with. Therefore, we will propose a brief list of questions to guide our initial look into what AI seems to have become in its latest incarnation.
AI in 2018 – Players and Questions – A short list to guide our inquiry
- Who are the players leading the current rebirth of AI and why?
- What does AI mean today, versus what it has meant in the past?
- What has changed, and is there anything new or different in what is being done?
- What kind of results are coming from AI research today, and what is lacking?
- How are the results that are being reported today being achieved?
- What are the risks inherent in the latest round of AI research and reporting?
- How can we understand and evaluate the claims of what AI is doing or is going to do, as reported by the popular press, or in other publications?
- Is AI a science, and if so, what kind of science is it?
The list could go on, and does, but we have listed a rather broad outline of issues to focus our attention for awhile. In this and a number of succeeding posts, we’ll consider these and other related questions, as appropriate. We won’t try to answer every question in one sitting, or in great depth. Otherwise we might write volumes devoted to some of the questions and issues listed above.
Disclaimer – We Consider Only What is Being Called AI
By way of disclaimer, we need to point out that good scientific research is being done in many areas related to what might be loosely called AI, but which is not being referred to as AI, today. We exclude these efforts from our current topic of discussion. In this series of posts, we are concerned only with what is being referred to in the media and publications as “AI”, or “artificial intelligence”, today. We are well aware of the claims that what is being called AI is not AI, but our interest is only in what is being called AI.
Let’s Begin the Discussion
To begin, let us go back in time to the days of the birth of the electronic digital computer and Cybernetics, and recall what those days were like. Recall that there were truly gifted people like Turing, von Neumann, Wiener, and many other intellectual giants and scientists who clearly understood what they needed to do and who blazed the way for the computer revolution to follow. They were associated with top level scientists and engineers from a large number of other scientific and technological disciplines, hoping to benefit, themselves, from the promise of the new computing technology.
As AI has made a number of reappearances in the more than six decades since its official birth in 1956, we seem to have seen a falling away from AI, of many of the kinds of people who were associated with the original computing and Cybernetics movements. This doesn’t necessarily mean that they, or their successors, aren’t still out there, carrying out their own areas of inquiry, as before. However, many of these people today have chosen not to claim to be associated with the current AI movement, by that name. This can perhaps be best understood in light of the recent history of the AI movement. Yet, on the other hand, there clearly is a certain group of people who appear to have assumed roles as leaders and innovators in what is being called AI today.
Speaking as a somewhat enlightened observer (only), and admittedly oversimplifying a complex subject, it seems that AI is now largely considered to be the almost exclusive domain of people who refer to themselves as computer scientists, or perhaps would-be computer scientists. These are not, by-and-large, mathematicians, physicists, physiologists, philosophers, or neuroscientists, although most of what they do is certainly and necessarily rooted in mathematics, and of course, computing, at some level. They are not typically neuro- or brain-specialists, and indeed some would admit that what they are doing has very little to do with the brain, which is too poorly understood to be a good guide at this time. However, it does serve as some kind of a model to be studied and emulated (dare I say copied?) to the extent that this might be possible.
There is another interesting aspect to the question of who the players are, since they are not just individuals working on their own or in small groups in labs or universities. In fact, they include many individuals working for some of our largest tech, social media, automobile, and retail companies. Indeed, these companies are providing a very competitive market for talent at this time, even at the expense of universities who can’t necessarily retain that talent. The question has been asked whether we have turned over some of our best efforts and talents in AI to social media companies, and even automobile makers, today?
Our first question might be to ask what motivates these people and companies to keep the AI fires burning? There are some obvious answers. Some (especially large corporations who are gambling on AI for future growth and profits) are motivated by money. Money in some way probably lies at the heart of many of the efforts being made today. If you are a young academic trying to build a career and a reputation, then funding for your research, and obtaining support for your graduate students, is very important. The term “AI”, or “artificial intelligence”, suggests a certain mystique that seems to attract investment money from many sources including both public and private. If you are an entrepreneur, then using the term AI to describe what you are doing, or going to do, is a great way to attract attention and all that can follow from that, today. So, the motivation is not really hard to find, and has not changed very much from the previous incarnations of AI. We recall that the great expenditure made by Japan for the Fifth Generation Project was expressly for the purpose of giving Japan a great economic advantage over the rest of the world. Indeed it might have done that if it had succeeded.
Our next question might be what are we really dealing with here and why so much emphasis on the term AI? Because of the potential promise of the mysterious term “artificial Intelligence” to bring really significant and important changes into our lives, it assumes an imagined importance that may or may not be justified, but which serves very useful purposes. And so there is interest in preserving and continuing the AI image as something very important and worth investing in. Now, recalling that there have been several AI winters gone by, and another one could always turn out to be the last, we note that the original goals for AI, as set forth in 1956 might actually never be fully realized in any reasonably near time period. Certainly they have not been fully realized, to date, as was promised. Hence, there appears to be a fervent interest, today, in just getting something, and perhaps anything that utilizes the computer, to appear to be doing something really significant or at least impressive, that can be called AI (and after all, there probably isn’t much related to the computer that couldn’t somehow be referred to as AI, in a very loose sense).
Most modern uses of the digital computer are not typically referred to as artificial intelligence today. Adaptive mathematical processes are not new. They have been understood, are readily available, and have been in use for a long time now, and they are more aptly named. They are routinely used in problem solving, signal and image processing, optimization, curve fitting, and automatic control, for example, to name a few familiar and truly important applications.
At a simple and perhaps somewhat naive level, these adaptive processes provide the basis for much of what is being called AI today. We see articles with titles like “what is being called AI is not AI” and we read reports by disillusioned students proclaiming words like “I thought this was supposed to be about knowledge”, and concerned that they are being encouraged to tweak results and make things look as earth-shaking as possible. Of course, those approaches can be important if you want to sustain the flow of money into what you are claiming to do.
Is it all hype? Perhaps not, but there is surely a certain element of that apparent in much of what we hear or read about. In fact, the AI movement of today almost seems reminiscent of the recent dot-com era. At that time, not so long ago, proclamations like “we don’t know exactly what we are going to do (on the Internet), but we are going to do something important” could lead to massive funding. This is, in fact, very nearly a quote from the dot-com years (1990s) and this kind of quote, while extreme, was about all that was required for a highly successful IPO, instantly enriching all who were involved (except perhaps the unwary investors). On the other hand, to be fair, we should keep in mind that not all of these new ventures failed, and one of them has become in fact, spectacularly successful.
If we accept the premise that the goal is to get something involving the computer to appear to be doing AI, (whatever that may mean) and to be pointing the way to greater things to come, then what this really boils down to is people tinkering with a new toy that surely can be used to do amazing things if we just hit on the right combination of ingredients. We are going to examine this idea in greater detail later. The child prodigy and mathematical genius Norbert Wiener talked and even warned about what he chose to call “gadget worshipers” in his book, God and Golem, Inc. In a few words, be careful what you wish for (there may be unexpected consequences)!
We don’t have to look very far to find the buzz words behind what the new toy can do, and they include terms like Machine Learning, Deep Learning, Facial Recognition, Object Detection, Style Transfer, Game Playing, and even Autonomous Driving, which is certainly one in which the risks must be taken very seriously.
The interested reader is further directed to in-depth scrutiny of Neural Networks currently being posted in Dialogues (on the menu bar), or click here.
What’s Next?
With this brief introduction to players and questions, we will continue our inquiry in the next post to appear soon. Thanks, as always, for joining us, and we’ll look forward to seeing you again as we delve ever more deeply into the world of AI in 2018, and continue our pursuit of an answer to the question, “Is It Deja Vu All Over Again?”.