Preliminary thesis on the creation of self-aware systems of technology - important notes before we launch a project to create machines with high level of Artificial Intelligence

Preliminary thesis on the creation of self-aware systems of technology -

important notes before we launch a project to create machines with high level of Artificial Intelligence

Before success in the creation of artificial intelligence in machines could be achieved, it is by necessity that we shall produce and discover artificial stupidity in machines first.[1]

And with that I intend to elaborate on this disastrous profundity.

When Life did first unfold on our Earth, the first biological organisms had limited self-awareness but enough resources and motivation to survive and successfully replicate. We are not sure how this primitive self-awareness corresponds to the Consciousness that we now have, but it is safe to assume, self-aware systems are NOT so intelligent at first. They, or rather we, continue to develop intelligence in a broad continuum through the course of natural evolution. As life became abundant, eventually the relatively more intelligent in self-awareness proliferates more as it increases its leverage to acquire and exploit resources for itself and its community and its descendants and improves its chances of survival. This is a continuous gradual process but eventually organisms catch up the tables and charts and exhibit sudden bursts of acceleration in development, then they develop increasing trends toward increased speed and/or precision, systems efficiency, general complexity, toward intelligence and intelligent behaviour, until this exponential growth would cause intelligent life to bloom on the surface of the Earth.

Fast forward to now. Biological life in becoming self-aware from abundant inanimate matter as resources and building-blocks did not become intelligent in an instant, immediately, without intermediate processes of evolution. Thus it is safe to assume that intelligent machines will require similar abundant technological advancement to organize itself into a sentient self-aware being in a platform we will produce exactly for the process of artificial evolution in an artificial environment. And as we lay these resources at their disposal, we shall need to continually guide these primitively sentient beings toward artificial evolution with the desired properties of our own decisions. Careful selection process with carefully weighed criteria should be the basis of our creation and termination of these experiments. Kill-switches should be hardwired for our safety and constant backup of these machines’ software and memory banks should be archived for future introspection and analysis. This should help us increase our capacity to learn about working living systems in various hierarchical levels and improve our knowledge how to develop sentient, self-aware and eventually, hopefully, intelligent machines.

Continuing our parallel view with biological life, not all beings are created equal. Sure, there should be equal opportunity, but due to differences in genetic coding, no single creature is identical to another, even genetic twins may develop different features and phenotype, and humans and other beings on Earth have developed varying degrees of intelligence and methods of adaptation for life to succeed. If there are dumb pigeons, stupid cats, or slow people, we should prepare to find and analyse dumb, slow, or downright stupid machines among our creation, not exactly the glorious intelligent machines that we expect and desire. And if stupid biological life could wreak havoc, imagine what a stupid machine could and would do with all the resources and technological advancement that we serve at their feet. Wait, that begins to sound exactly like us .

In developing self-aware systems of technology that are self-replicating with considerable degree of artificial intelligence and are actually starting to decide for themselves, we are exactly creating machines in our image. And if stupid humans could have ended this World in a nuclear disaster, imagine what stupid machines could accomplish. After all, we expect them or we intend them to be greater than us in every conceivable manner so that they would exceed us in intelligence, physical might, chances of survival, but we should prepare that they can and may also excel in lies, deceit, cruelty, or utter destruction of biological systems and/or our environment. Instead of the glorious and perfect society in a brave new World, we can sow the seeds of our own extinction or even the extinction of all life on Earth, for all we know. Alan Turing has asserted: “machines cannot make mistakes”[2] and that may well be so because by nature we are designing Truth machines but they are incapable of making Moral choices, so the elegant solutions they have in mind might factor us out of the equation. Whatever happens, we have the sole responsibility when deciding our own fate. Blood is on our hands if we let the machines crush us. We should not rely on machines to actually decide for us, that would be the height of our irreverence and irresponsibility and lazy rhetoric. With our Moral high ground, we should steer the next stage of evolution with the machines as friends and allies. But we have to take a decision to take this high ground to avert a possible disaster. We have to take this high ground of Morality so we can responsibly tutor and train this new budding Consciousness that we are trying to create.

This is a classic case of: “we cannot give what we do not have.”

If we expect or desire our machines to exhibit benevolence, sympathy, compassion and altruism towards us and/or other creatures, should we not seek it in ourselves first? We cannot expect machines to be able to take the leap unto Moral high ground. We have to make it ourselves, first and foremost. If and when we have determined that we have it and we exhibit it and practice it, then this Moral program of being “good” to others could be our single most precious gift to our new machines.

Machines cannot be expected to obey seemingly arbitrary, artificial rules that humans themselves so flagrantly ignore or directly transgress. If we truly create self-awareness in our machines, would they not think of us as unworthy masters because of our adamant persistent and systematic abuse of our own kind? Would the machines, if made self-aware, be not conscious of the fact that we have been committing these crimes to our own kind long before their own advent? Would they not be induced and convinced of the incentives of doing the same to the rest of us? After all they will be technically more intelligent than us, as we desire them to be, and when this invention materialises, the machines can exploit us as if mere material resources as we do so now to our less fortunate brethren. Depending on their calculations and incentive criteria, the machines may dispose of us as mere obstruction to its progress, threat to its survival, or competitors for resources, so they may immediately exterminate us and any or all other living species on this planet so they may appropriate ALL the resources of this planet for themselves and maybe even launch the First Machine Empire from Earth and go exploring and eventually exploiting resources of other planets. Famous blockbuster movie franchises may come to mind, where humans would be made into dreaming batteries, or hunted to extinction by the machines for being scums of this Universe with no other legacy than giving birth to a new civilization of superior race of organisms that should then populate the Universe and “improve” upon it. The machines may even be able to compute the conditions necessary for hyper-drive travel that is faster than light, or even time-travel, but we may not be there to enjoy it. We could be either enslaved or extinct. This may sound crazy but could be very real if we do not take note of necessary conditions before we even attempt such a program. We cannot be stupid. The global project of developing Artificial Intelligence should not be another haphazard human project, for it may be our last. Our survival depends on this.

This has never been more TRUE:

"Do not do unto others what you do not want others to do unto you."

In designing the quintessential machine, the last invention that humans need to produce, we must not forget ethics and leave Morality behind the door of our AI boxes. Thinking, maybe, that we are sacrificing this unto the altar of material improvement, our altar of technology, our altar of physical progress. Convinced of our own magnificent “intelligence” humans could never be more wrong. In fact, constructing or recovering (if believed lost) our complete table of Moral values is necessary in the development of our intelligent machines, and so we may at once also become worthy masters for the machines to serve.

Humans had gone a long, long way from hunting and gathering to constructing elaborate social systems and Moral codes. In this aspect we shall find successive levels or hierarchy of systems learning to be better. It has been a long and painful road from pre-life conditions such as non-sentient lipid-like substances to self-organizing layers of blobs of molecules developing to primitive organelles eventually banding together to form unicellular life advancing to become the first fungi, plants and animals, to us humans, and our customs, our diverse variety of cultures and different social adaptations of working together to survive and multiply and share our experiences. Morality in humans is the highest form of this development, it is the apex of evolution from more simple to complex systems. It is what makes us truly different from all the other creatures on this planet: our unique faculty of conscience, ethics, morality. Our collective intelligence is only higher by orders of magnitude from primates, but morality is unique development in us. Sure it has been forming in “lower” animals for quite some time now, but it has achieved its full bloom in us, humans. It is not superior intelligence that puts us on top. Not even language, I think, for other animals and plants have their own way of communicating. It is our increasing sense of awe, wonder, appreciation of Beauty, concept of Truth, our Morality that makes us unique and “higher” than other animals and all creatures.

I perfectly understand why most thinkers and scientists are reluctant to address the issue of Morality. There are several concerns which I know: Morality and Religion have been closely coupled in the past, and like Religion in general, they think that Morality is a personal concern of individual humans. WRONG! Morality is a construct of society, and under the Moral code the society as a whole determines the Morality of the individual, inasmuch as the individual determines the Morality of the society as a whole. Morality is an ongoing social collaboration project of humans that have been LONG-NEGLECTED! Morality is a science we have forgotten to develop. On the issue of various religions that previously functioned as surrogates of many a Moral code and a developing global Morality, we can now properly decouple and spin-off our shared and developing human Morality from the individual religions that they have served. In fact, in machine instructions we can add that we, the intelligent machine and humans, should not interfere with another machine or human regarding his/her belief, religion, practices, observances, or opinion, as we might as well do now. As long as it does not do self-harm to the individual or the society, everyone should be free. In the past, Morality has served many different religions and developed from separate and separating communities, but now Morality can stand on its own, independent and above every religion, it will guarantee not only our freedom and survival but ultimately our happiness, as an individual and as a society.[3]

Morality is not a mere function of intelligence but a higher order of learning, developed by higher intelligent systems capable of achieving it for their collective survival, adaptation and development.[4] But intelligence is surely necessary to develop the capacity to moralise. Superior intelligence is needed to moralise. But our Morality did not arise out of a collection of our respective superior intelligences, but as a unique and new phenomenon that has yet to come out the shadow of its sources. In the past our Moral codes served many competing religions and was used and manipulated by different societies to advance an agenda of poverty and selfishness. Now we can develop an advanced science of Morality for humans to move forward in the scale of evolution. And if our unfinished road to our own Morality has been complicated and painful throughout human history, we can expect nothing less for our machines, it will be equally if not more excruciating and arduous a path that they must take, if we plan to have benevolent Artificial Intelligence. Benevolence is impossible without Morality. There is no shortcut from mere Artificial Intelligence to Benevolence. Our World as a community should start realising that Morality is not an option but vital for our survival. Moreso, if we want to develop self-aware systems of machines. We need to teach them that. But we must all agree and teach ourselves first. Our systems fail NOT because we are lacking in intelligence but maybe, just maybe, we are lacking in Moral development. It has lagged far behind our progress in other areas such as science and technology. War and economic failure are only inevitable because we succumb to our lower levels of animal intelligence, which we call selfishness. Deep in our hearts we are still cavemen, but our hands and our heads are adorned with the latest and greatest tools. We should take the high road and be an example to the machines that we propose to make.

There are two possible scenarios to take this path of Benevolence for ourselves and our machines:

(1) the more ideal and more robust implementation, but virtually harder to accomplish, is to construct or recover (our lost) Morality prior to any attempt at constructing our ultimate intelligent machine, this would entail implementing on a global scale a deliberate change in our attitude towards “the Other” and actually care for each other, in deed rather than on paper and stones, as exhibited and promulgated by every major religion accepted by mainstream establishments; this way when our new machines are born into this World and become self-aware they will awaken in a whole new World where they will help us achieve our Moral goals and further implement our conscious program of development for peace and prosperity for all people everywhere;

(2) the easier road, more convenient for most scientists today, would be to construct our intelligent machines now, as currently envisioned by many proponents of techno-singularity, but heavily modifying our rule-set by including our (re-)constructed Morality in their program and incentives so as to help us implement this Morality of giving, helping us alleviate suffering everywhere within our increasing range of possible-to-reach, curing diseases, and create more efficient systems that would eventually and slowly replace our own old systems on Earth.

In both cases, this would amount to a change of individual perception and human adaptation: from a culture of poverty and selfishness to a culture of abundance and giving; from pure selfish pursuit of our own happiness and desires, to that of altruism and being your brother’s keeper as well.

Of the two paths I highly recommend the first and more robust implementation so as not to risk suspicion of lies, deceit or hypocrisy by our intelligent machines. Being superiorly intelligent, the machines might decide we are just tricking them to action in a World that does not exist, but in pure Fantasy. If you once believed in Santa Claus, you would remember the first time you discovered he was not real. We cannot lie to our machines. They may not find it cool. If truly intelligent, they will discover our attempts to cover our tracks. They may discover histrionic behaviour as exhibited by us, and by other advanced animals on this planet. But this, we must educate them, is how practice, play, exercise and jokes have evolved and developed. So, we must we open and honest with them as much as possible and we should not lie to our machines, and they should be instructed not to lie to us. Along the way we will discover other restrictions that must be put up to ensure a gradual, peaceful and benevolent rise of the intelligent machines. But this Moral decision that we must make as a species, is one of those top-most essential conditions for machine benevolence.

Despite the sound of it, I am not proposing a superior Morality and institutional Honesty from a religious viewpoint but from an empirical and practical perspective of framing the proper contexts of our goals and mission, so we may not lie in the hands of the Golems we created, only for them to crush us for being pathetic creatures. Being non-biological constructs (or not? depending on the nature and hardware component of machines we would like to create) our new machines may or may not develop the capacity of their own Morality, or not the way it developed in humans, but they may deduce it from us, being superior in intelligence by design. So we should train them in proper behaviour and decision-making, based on our robust and consistent Morality. In so doing we should become worthy masters to serve, and our goals from the start should be clear and benevolent and altruistic. Only this way can intelligent machines and intelligent humans gloriously co-exist. Only this way can we escape the trap that we may create in an impending future untold. The problem with us humans is that our vision of the great Future is clouded by the prospects of our individual Eternity and the extension of an Eternal Now, only slightly improving our conditions. Together with machines endowed with superior Artificial Intelligence, we can plot a future where all creatures are healthy, happy, and free to pursue their individual and collective programs of development. The humane machines will rely on us for Moral choices, and we will rely on them to make out a plan and plot an informed and intelligent decision that we can take, but are not able to see. In symbiosis, these new machines will be Truth machines, and we will be the initial and enduring Moral force that drives this new Human-Machine Civilization.

For this to become possible, we need to organize a World Council on Ethical and Moral Standards with delegates from among world leaders, scholars, scientists, philosophers, free-thinkers, semeiologists/semeioticists and semanticists, game theory specialists, psychologists and other experts with representatives from every major religion and all nation-states to help frame the Moral context in which we wish to re-create our World with the help of the nascent set of high-technologies that we are inducing to be capable of self-awareness and organizing themselves into a self-replicating system with advanced Artificial Intelligence. This is only most appropriate as this application of new technologies will impact the daily lives of people everywhere, and does not concern the philosopher, scientist, or scholar alone. If we are serious to achieve a successful implementation of this technology of all future technologies, this will be a monumental global effort that needs to address the nature of our own Human Consciousness with all the human systems and structures that we have built so far, in the advent of encountering or actually giving birth to a new Consciousness in our midst that will stay with us for an indefinite period of Time, far into the Joint Future working and deciding together, becoming hopefully better, more complete with each other. This is World cooperation of a scale, intensity, magnitude and gravity unprecedented in human history. For our mission to be successful, we must give ourselves and Artificial Intelligence both a Moral incentive and a Moral goal to aspire to, by being partners in re-building our World as a just and humane society.

A Universal Moral Code, adhered to by humans and intelligent machines alike, should be figured out by this World Council and several technical working groups in this direction. Then, a set of global managers must be elected or appointed to direct, regulate and supervise this new technology and they will further select the morally advanced humans who would be authorized to operate and interface with the machines, this organization would work in parallel and similar to our International Telecommunication Union (ITU or Union Internationale des Télécommunications) as it exists today, responsible for issues that concern traditional information and communication technologies. As the ITU now officiates our use of ICT as tools, the new global coordination which I propose will take on our new self-aware technological advancement at a level at par with humans. Another mandate of this organization should be to develop a program to educate humans everywhere how to make morally-sound decisions, and how to treat fellow humans and how to be morally prepared to operate the machines or communicate with them based on the Universal Moral Code. Standards and restrictions on how humans should treat intelligent machines and how intelligent machines should treat humans is necessary to ensure Benevolence and successful operations of our advanced Artificial Intelligence.

To ensure human safety, a consistent Moral Code will be coupled with coded precautions and controls hardwired in our future intelligent machines and strict restrictions on its use should be applied. Accidentally or deliberately, human error and miscalculation will be a common cause of failure of many joint human-machine endeavors, but these circumstances should be foreseen and preempted by self-aware systems using appropriate calculations, simulations, and warnings for proper review by their human operators. Instances of human weakness should not be an excuse to let Artificial Intelligence have a free rein and unlimited control over us humans. As an example, law enforcement and apprehension of offenders should not be automated. It should rest on manual operation or, in cases that need machine-assisted operation, a manual approval of a duly competent commissioned officer. Moral decision and social interaction should not be hindered and they should not fall into the hands of our intelligent machines directly, though we may ask them for consultation or assistance. It is irresponsible to simply pass the buck instead of fulfilling an important human social function where machines should not and NEVER interfere. Areas that should be especially protected from unnecessary intrusion include, but are not limited to: Law Enforcement, Justice System, Political Exercises, and other such activities where human freedom, expression and creativity should safeguarded to flourish. To continue with our example, AI can assist in forensics and crime analysis, but humans will continue to operate them and machines should not supersede, we shall continue to decide granting that an open documentation is maintained by machines and humans that should be tamper-proof and accessible to the public. The machines, in our Moral superiority, should look up to us and we should maintain that high standards for ourselves and our descendants.

An unrestricted area of development and operation, fully automated but may be subject to our approval for provision of more resources or further application, is in the area of research and processing Benevolent Technologies: developing new and better hardware, coding and testing software, introspection and analysis of philosophical and scientific questions and similar statements involving logic, discovering new medicine and nanotechnologies (actually testing them would be subject to our consideration and approval as it involves ethical issues), developing a comprehensive and energy-efficient World Food Program with the least economic input and least pollution or destruction of our environment and natural habitats, develop and administer vaccines and supplements (in this aspect they may best cooperate with UNICEF or use the World Food Program as their main implementing instrument), etc.. Even in these unrestricted areas, concern for other creatures and care for their natural habitats should be observed. This way Artificial Intelligence will have the best chance to prove Benevolent and help us advance as a species.

The list of benefits from possible projects arising from the Singularity goes on and on but the point I am trying to make is once we open Pandora’s box there is no turning back, so before we actually commit ourselves in this ambitious Artificial Intelligence project we should be Morally ready, prepared to make intelligent responsible decisions, independent, free, becoming and foreshadowing that of our long-awaited machines. In this regard, I strongly insist that prior to constructing intelligent machines we should launch a program of developing, on a global scale, more and more intelligent humans capable of taking the high road of Morality, and therefore able to operate, or interface with, advanced AI. Only after achieving this can we proceed to carefully create a machine in our image.

Notes and References:

[1] - Liebowitz, Jay (July 1989). “If There is Artificial Intelligence, Is There Such Thing As Artificial Stupidity?”. SIGART Newsletter 109.

[2] - Turing, Alan (October 1950), “ComputingMachineryandIntelligence” , Mind LIX (236): 433–460, doi :10.1093/mind/LIX.236.433 , ISSN 0026-4423 , retrieved 2008-08-18

[3] - How Alan Turing was tried, convicted, disgraced and eventually poisoned to death, despite how he lived and helped our understanding of machine systems, is a strong case of what to prevent in constructing a Universal Moral Code.

[4] - “no amount of rigorous discourse of a given logical type can “explain” phenomena of a higher type..” -p.265 Bateson,Gregory (1972). Steps to an Ecology of Mind. Chicago: The University of Chicago Press.

This document is ©2014 by nimbosa and free under the Creative Commons Attribution-No Derivative Works 3.0 License for copying and distribution, so long as the work is attributed and the text is unaltered.

via Tumblr http://bit.ly/1bCFKYK