In my previous note, AI, in a strange way, was the first and least important subject for my personal Christmas meditation. It was first because there was so much disconnected discussion over the entire sphere of human (perhaps with some alien intelligence influence as well, who knows) discourse. It was last, because my understanding was dated, and, the discussions on offer, seemed, more glitter than substance. I certainly got caught up in it. Related topics, from my perspective, ran the gamut of:
Greed driven, how do we use AI to make a buck without effort, even more rent seeking, types of discussions.
Power driven, how to control the stinking masses without risking ourselves, desires to become the masters of the world.
Human exceptionalism driven, don’t touch my divine carcass. Leave me as I am, no more, no less. My soul must within this long suffering flesh sack remain. AI is a temptation from the infernal beast. (You might ask, where is AI on this point, it’s there, in all its glory, I will expand on this, the point below touches on this)
Is AI, the holy grail? Have we gods become? Have we, as we are, become irrelevant? Worried about loosing yourself within the AI world, not to worry, word salad will save you. AI (strange, I thought, if an acronym was to be used, it should be JC, maybe another alphabet) is the second coming, you will all be resurrected into the “perfect” you, and, if that you is less than expected, don’t worry, the other you was just the imperfect you, this is the perfect you, whether you understand it or not, and perhaps, whether you want it or not.
Are we pulling our hairs out, on this AI topic, for nothing? Is AI, in its current form, to immature to be fruitfully discussed ? Certainly, within the context of some of the previous points, it certainly seem to be (Spoiler, I currently am firmly in this camp). We certainly should have the previous conversations, however, like many of the topics we are told to be very afraid of, exist mostly in the data-free conversation realm. The conversations should be placed in our mental receptacle labeled, interesting verbiage, use with caution.
Before we can even have a sense of what we might be talking about, we should perhaps distinguish the difference between AI (as appose to ai) and machine learning (ML).
To me, AI is the catch all acronym for anything that makes an algorithm seem human or super human (generally, never spending the time to explain what being human means, in the first place).
ML, on the other hand, (again only to me, if you want another viewpoint, this might be more to your liking) is mostly regression type analysis. What do I mean “regression type analysis”? I mean, when looking at a whole bunch (or a small bunch) of different types of information, we might attempt to determine if there exists relationships between the different types of data. The purpose of this exercise would be to use one set of data to predict or guess at the value of the others. There are innumerable methods to make these associations. The types of data are often referred to as variables. So, we could say, we would like to predict one set of variables given data of another set of variables. I think, an example will, hopefully, clarify this paragraph.
The simplest example (that I am aware of) is the case of two variables, we proceed to plot both of these, a piece of paper will do, specially if a computer has become a bit too threatening, with AI lurking in the background. It is assumed, we have a dataset consisting of both variables in pairs say: (x0,y0), (x1,y1), … , (xN,yN). Here, both the x’s and y’s have been labeled with a number from 0 to N. In other words, we have N pairs of the variables x and y. So, I shall plot an example, the result is shown below.
The plot above has 10 points in reddish colors. Each point is one (x,y) variable pair. To me, and probably most people, the points appear lie along a diagonal. We have just accomplished most of what ML does. The simplest thing to do would be to draw a line through these points and say, something like, the line is in some way relating the x values to the y values via the proxy of the line; therefore, if we care about any other x and we want an “estimate” of what the associated y might be. We just go to the line and read the new/predicted y value. The plot below contains one possible line among infinite possible ones we could have chosen.
The plot above shows our chosen (best fit in some sense) line in blue. If we like this line, for what ever reason, or, no reason at all, we can pick any x value and read off the corresponding y value on the line. This exercise is represented by the blue dot on the upper right of the plot. This was certainly not a point from the original dataset. We have just done the magic of predicting a y value given a training dataset. Are you underwhelmed? If you are, lets move over to the computer world and look at this process there, and, see if you are still underwhelmed afterwards.
This is, on a fundamental level, all the magic of ML, which, a computer performs for you. The x values of the graph above is replaced with a bunch of different x’s, and, the y values are replaced with with a bunch of y’s. We call the bunch of x’s the input (computer-ise for things we jam into the computer black box) and the y’s the desired output. Let the computer form its computer controlled relationships lookup (like the line we “discovered” in the simple example above), and viola, for any other observations in terms of x’s, the computer spits out y’s. The differences are:
the large number of inputs a computer can deal with,
which also applies to the output,
the large amount of correlations between the inputs and outputs (in my minds eye, I see an ocean of plots like the one simple one we generated),
the huge amount of control of these correlations and the huge amounts of control we could abdicate to the computer.
This is were, the boggling of the mind comes in. The boggled mind tends to see magic where none exists.
Here, is one persons interpretation of “art” as assisted by ML algorithms.
The word intelligence attributed to the algorithms that helps in his creative process is surprising. This view of intelligence is one of the basic questions I am (and I assume many are) questioning.
I never thought that when artists started throwing buckets of paint on a canvas as something controlled by some alien intelligence. I saw it as creative individuals using a different medium to express their internal artistic inspiration. A sharing of this inspiration, in physical form, with the wider world. I truly find it difficult for the word AI, more appropriately ML, to be associated with this. But it is done over and over again.
The problem, as I see it, is similar to attributing abilities to a human sperm or egg, of a non-existent full grown human. AI (which still, in my mind, needs to be better defined, if, it is to be a proxy of being human or superhuman) is still in a very rudimentary stage, to attribute too much super human or god like attributes to it, is unwarranted and can be used in very dangerous ways. It will be tempting, for some, to use computer algorithms to control your life in very invasive ways. To list a few:
AI surveillance, is not just to protect you from the bad guys. It can and has been used to identify and curtail many individuals that even dare to question, express disagreement or expose what is perceived to abuse of power of those that are doing the abusing. Which is quite understandable.
AI control in war. How can the life and death decision be given over to rudimentary algorithms. I am quite sure they are not, those that are really in control will abuse their power and blame any fallout on the computer.
AI control in healthcare. This could be good, but, it can also be another form of totalitarian control. If you trust those who have their greedy fingers all over the AI buttons, there is nothing to worry about, medical utopia is around the corner. It is similar to seeing nuclear energy as a potential source of abundant energy, or, a threat to life on earth via the hydrogen bombs, some have grown to love. Are these in the hands of a similar set of individuals?
This list goes on and on, every aspect of your life is being offered up to the AI wizards behind the curtain. Those who pay the wizards call the tune. Who is in a position to do this? How much better is it now, that, they are getting more and more access to the levers of power? How transparent have they been, when rolling out these processes?
As I see it, AI is in its embryonic stage. The impact of computers and ML is incontrovertible; however, ML is not AI. ML does not provide an unbiased decision making process. ML is being given the aspect of intelligence, which, it certainly does not deserve. The great danger, as I see it, is to abdicate our own thinking and understanding (I am making the leap, that, we do actually have some real intellegence). AI, as an aspirational goal for computing, might be OK, if we keep a clear eye on it, but, it is too early to know where it will go. The equating of ML with AI is ludicrous. To say that ML is a vehicle towards AI, is again fraught with ambiguity. This can be compared to, viewing a rickshaw as a vehicle towards manned space flight to Pluto. Potentially true, but, who cares. First I need a strong definition of what AI is. Secondly, we need to control those who would want to abuse the crude technologies we are currently stuck with. These elitists don’t need to understand anything, they only need to know how to put spin on a magical seeming process to bamboozle the masses towards more totalitarian control.
On that bright note. I shall conclude by saying this cannot be concluded by me, perhaps by anyone. ML has much promise, but, as usual it can be used to the determent of the many for the benefit of the few. The second possibility, at this point, seems more likely. The key takeaway is, anything less understood by the many can, and has been, used against us. We all need to educated ourselves. The elitist will not educate, they prefer to propagandize us, there is no profitable rent in an educated public. Feed them with invisible threats that is just outside of easy understanding of the many. Use this lack of understanding as a means of instilling fear. Use the fear to usher in elitist controlled savior figures. Use these figures to further enslave the many and enrich and empower the elitist. This is a well warn path, one we must divert from or humanity may well cease to exist under the cruel mechanization of the elitists and their puppets. An interesting person, Dr. Patrick Morre, who talks about these invisible threats in an eloquent and accessible way is well worth listening to. I certainly am looking forward to reading his book on this very subject.
I will have to keep my mind open, mouth, ear and eyes uncovered on this subject with the hope of approaching some truth. Hey, your not a monkey stop covering up. LOL.
Related articles from after this note was written: