To humanize the AI, or to dehumanize the man?
It’s Saturday evening and I’m having a conversation with Jacek Santorski and Kamila Goryszewska. It’s good, easy-going and with style. I say good because we listen to each other carefully, don’t interrupt ourselves and inspire each other. Easy-going, because without an imposed structure, topics come and go freely. We are opening the door of the mind to further independent thinking, rather than closing the abundance of subjects. Henceforth, some of my individual, silent reflections:
One of these subjects was artificial intelligence (AI); to be specific, quite a few separate subjects, including the issue of AI empathy. I already wrote a separate article on this one, so I was prepared to share some thoughts with my interlocutors. Returning to this issue in the discussion inspired me to reflect further on the directions of AI development. Taking into account the criterion of "man as a reference point", 3 such directions came to my mind:
1. Humanizing the AI, or developing it like a human being
2. Dehumanizing the human being, or cybernetic enhancement of functions
3. AI combining the advantages of people and machines into a new quality
The first direction was suggested by Jacek Santorski. He considered how to make a (ro)bot feel empathy. He concluded that empathy is a purely human phenomenon, so it may deliberately deem the (ro)bot imperfect, making systematic mistakes and weird decisions. Just like a human being who can learn, but to a limited extent; who has equally limited abilities to collect and process information, so in the process of evolution developed ways of so-called economics of thinking (the effect of thinking as little as possible so that the system does not overheat). Only then there is a chance for the (ro)bot to develop such phenomena as empathy, and as a consequence a chance for AI solutions to enter smoothly into the world of interaction with people. Well, it takes one to know one…
It is a very interesting thought; however, personally, I think it’s strongly unconvincing. Since we already have nearly eight billion individually imperfect beings (and that number is rapidly growing), why create something that will only recreate and imitate their imperfections? And make the same mistakes, because the initial thesis that "empathy is a purely human phenomenon" is not defendable. This, as I’ve mentioned, I discuss in a different article. Such a humanization of AI is rather an expression of longing for the creation based on "the human model" (because "a human being is the measure of all things", with the multitude of meanings behind this statement) and fears of where the other two scenarios can lead us. The latter probably even more so...
A side note: of course, AI will also learn and this learning process is already the foundation of its current prenatal development phase. It's just that once it’s learned something, it will know it forever, so why implant distractors of purely human origin?
A side note number two: speaking about distractors of human origin, I do not mean incorporating ethical threads to AI solutions. This is another issue.
The second direction – dehumanization of humans, which means a permanent cybernetic strengthening of human body functions. The starting point is: we implant non-biological elements enhancing the existing purely biological possibilities. The key word here is enhancing, that is, rising to a previously unavailable level. The term augmentation is slowly beginning to make a career, let’s see if it catches. For now, it’s doing great in computer games, such as the extremely successful Deus Ex series, where the article's illustration comes from. It is possible that it will stay in our language for good. I honestly think that this is an inevitable direction. It will start for obvious reasons; actually, it has already begun. Isn’t it good to save lives or restore irreversibly lost bodily functions? Of course it is good and it should be applauded. Think heart transplantation and ever-perfected artificial implants, especially when the donors are scarce. Limb prostheses, endoprosthesis, communication systems for the paralyzed, e.g. operated by brain impulses. Systems that allow for partial regain of eyesight. Today they’re all in their infancy, like first cars at the turn of the century. It starts with something basic that barely simulates the functions of the replaced organ, inferior to the original body part. But let's compare the first car with the ones produced today, which are still only a stage of the process of further designs... The first prosthesis of a hand was a curved metal hook, and the leg prosthesis was a wooden stake. Today, they are beginning to be quite complex mechanisms, the improvement of which practically has no end, until the prototype functions are exceeded. It's an inevitable process.
Let me remind you of the discussion whether the South African runner Oscar Pistorius should be allowed to compete because of the prosthesis of both legs. The considered problem was whether these prostheses give him an advantage over other runners. And yet his prostheses were a purely mechanical solution, still without any controls, let alone connection with the nervous system/the brain. But the work on such solutions is underway. The question is no longer whether it is possible, but when and if there are good (and cheaper) alternatives.
If we implant an artificial heart, why not add the utilities that monitor the blood composition, heart rate and other functions? The technology will enable this. Well, if it can save life at the beginning, let it save life. Over time, instead of saving those in critical condition, we will ask: what about prolonging the lives of people, whose heart is on the verge of usability, and the rest of the body, including the brain? Will we deny them the right to live? Of course, we won’t. So we will implant the heart that improves the original. And from this point on we’re only a step away from the next question - why wait for an accident or illness, if you can improve the body... just in case?! First, those statistically prone to disease; then, anyone who wants and has the funds for that; and then... everyone.
Well, we're actually already doing this by vaccinating, taking medicine and using a lot of other treatments (up to aesthetic medicine). If today's endoprosthesis, hand prosthesis or artificial hearts are OK, then where is the "not OK" line drawn? There isn’t one. It's just our thinking, different to every single one of us, which means, subjective. For a man of the middle ages, a pacemaker would probably be the same shock as for some of us an artificial heart with dozens of additional functions that are not in the original.
A similar train of thought: we will start with systems that enhance the eyesight of chronically ill people; and since we have partially restored some of the lost functions, why not expand them, for example, with the analysis of wavelength inaccessible to "ordinary" eyes. Same with hearing; first, we save and restore, then ask, why not expand it, if the technology allows it? And brain functions? Here a lot can happen...
For those who are interested in it, I recommend further reading by a Harvard Professor, Michael J. Sandel The Case Against Perfection: Ethics in the Age of Genetic Engineering. It is worth reading, although my final conclusions are different. Michael Sandel is trying to set a limit to the process of cybernetic and genetic interference in the human body. In my opinion, it's a fight you can’t win. You cannot limit an unavoidable process, originating from an obvious necessity that is saving life. And then, a step further, etc. One can and should discuss how to do it, simply not to follow the Frankenstein's path, that is, shortcuts. For the whims of tyrants or millionaires.
A side note: I am convinced that shortcuts are already being analyzed somewhere... Please take a look into my previous article "Ethics in business - Game Theory". You won’t change the laws of physics...
The third direction - AI combines advantages of people and machines into a new quality. It's starting to happen right before our eyes. As I’ve said above, the infancy, the embryonic state, we are just at the door. But this door has already been opened, because we are moving from the phase of intellectual discussions (philosophers and scientists) to tangible technological solutions (engineers). And we can only see what... allows to be seen. As this particular direction is the subject of my reflections in the emerging series of articles, there is no need to elaborate on it here. Thus, I’m only signaling it and inviting you to read the others, which I will systematically share. The key word here is "advantages"...