SuttaCentral

Buddhism & Robots: a rarely intersecting intersection


#1

On the subject of the rarely intersecting intersection of Buddhism and robots of late, if others would be interested further in intersections of such manner, the Korean film Doomsday Book (인류멸망보고서/“Report on the Destruction of Mankind”) has a very compelling and fascinating section about an AI installed at a monastery who claims to be enlightened after overhearing and learning Buddhist discourses, having the Dhamma indirectly expounded to “him”. I am still not sure if it was intended as perhaps mildly critical of some framings of Buddhahood or not, its just one of the “out there” ideas entertained in science fiction. An interesting and rare intersection of Buddhism and sci-fi, though.

I know I am without desire or compulsion, in the past, present, and future. I have learned that this is as the teachings of Buddha.

Humans, for what do you fear? Compulsion, desire, good deeds or bad, enlightenment and oblivion. The world this robot has seen through its bare nature was already complete, in and of itself.

-the “Heavenly Machine” (Kim Jee-woon et al., Doomsday Book)

Thoughts?


In Mahayana, Buddha-nature is the Universe Itself
#2

I will go first. Obviously we are no where near the ability to create a machine with any kind of intelligence that a biological thing would have. I suppose the idea is interesting on the level of sheer proliferation. Indeed, if something can formulate the cogent argument: “I have heard the Dhamma, I have understood the Dhamma, the situations the Buddha describes apply to me.” That has to be at least looked at. After all, there is no reason why, in some fantastical future, a fantasmagorically augmented humanity may bring itself to invent opportunities for lower (or equal?) intelligences to flourish.

But what if that isn’t the case with the “being” that states: “I have heard the Dhamma, I have understood the Dhamma.”

The argument is predicated on the “bare nature” of the robot, a non-discriminating filter for sensory input, which itself is “unconditioned” by what it experiences, not “feeling” anything pleasurable, nor “feeling” anything displeasurable. I will admit, that in my limited understanding, there are a few ways in which “linguistically alone” some things that can be said of mindsets conducive to progress on the path that resemble the “bare nature” of the mindset of the robot. The difference being actual sentience. I would never say that such a thing were the case, that this “bare nature” of the robot were the same thing, but at the same time, the similarity is a bit uncanny for me. Perhaps my own eccentricities are to blame. Perhaps others can see where I am coming from in this comparison. Obviously sentience is what matters, but we can’t even decide what that “is” on a “proveable” level. Patients who are “brain dead” can awaken spontaneously and have been aware of their surroundings the entire time, with no indication of awareness when scientifically studied. One reading of anattā argues that we are actually much more like the robot than we are comfortable with. I am not necessarily persuaded by it, though.

The presumed insentience of the robot is what makes the statement, arguably, most interesting, as I have before encountered descriptions of Nibbāna/Nibbāna-in-this-life, which I have always critiqued a bit internally, that are much like this. That is to say, they end up positing a Buddha that is much like the insentient robot, which I would ultimately not refer to as possessing the same “direct knowing”, despite that the robot does, indeed, actually have a kind of “direct knowing”, an access into a sort of “things-as-they-are”.


#3

This matter is not exclusively science fiction, nor is it new in fact, the idea of deliverance of consciousness through technological means has been around since the 80s, and is an ongoing subject of serious scientific research and investigation. Scientists working on this have produced a voluminous amount of data so far, some purely technical, some philosophic and even dharmic, including debates and differences of views regarding the old topic of the nature of consciousness, imagination, emotion, memory, and intuition, etc. Some of these debates are very interesting for any one intent on understanding the mind, involving the views of such renowned mathematicians and physicists as Roger Penrose. This field of scientific study is now growing even further given the recent and ongoing advance in neuroscience.

Some of the issues addressed in their work is directly relevant to dhamma, such as the effects of overcoming illness and again, even death, the qualitative transformation of the relationship between mind and body, the impact of strong AI on human cognition and mental experience, etc.

In case you don’t already know these, here are some useful links if you are interested:

:robot:


#4

Indeed, the interesting thing is the robot’s claim to have insight into the Buddha’s teaching is based, not on its really having “learned” the Dhamma as a freely thinking being, but rather, by having experienced it through a “bare nature”, or an absent nature. The idea being that this bare nature sees things “as-they-are”.

Very interesting. Also very superfluous.


#5

Hey, thanks for the info, it’s a fascinating subject. FYI, the last thing I did before starting my translation project was a panel with a bunch of futurists, including:

  • a robotics engineer
  • a professor who was growing a set of neurons in a petri dish and teaching them to play music
  • a legally recognized cyborg with a brain implant allowing him to communicate with satellites
  • a performance artist who, among other things, did experiments where his body was under remote control

Needless to say, it was awesome.

I think the frontiers of AI and Buddhism is very promising, and the contribution of Buddhism is sorely needed. As to the root question of whether computers can be sentient, in my view, this problem is still too often misconstrued following Turing’s question: “can machines think?” In the dhamma, of course, the key question is not thought, but consciousness or sentience. So long as this distinction, and the dependence of thought on consciousness, is not clear, there’ll be no way of answering the question.

In general, I would say that there’s no reason in theory why a machine should not be sentient. But I don’t think current computing technology is on track to achieve this. Something radically different is needed; perhaps the only technology on the horizon that might fit the bill is quantum computers. Because they operate in shades of grey, not just binaries, I think they might fit the bill of serving as a vehicle for consciousness. But that’s purely speculative, of course.

The idea that an artificial consciousness would experience things through a “bare awareness” is a fascinating idea, but I think wrong-headed from both a technological and a psychological point of view.

From a technological point of view, the idea that code is neutral is a fantasy of the technologists, an ideology that coders hang on to with a rather suspicious degree of emotional investment. But the reality is that code is the product of desire. Literally every element of code in every computer program is there because someone wanted it there. And this is why, as studies are showing more and more, programs show biases of sexism and racism and all the rest, no less than humans do.

As for the psychology, the idea of an unfiltered bare awareness is not really accurate, as all consciousness is conditioned. But insofar as it can be achieved, through an equanimous response to stimuli, it is not by failing to achieve a partial consciousness, but by growing past it. That is to say, our ordinary consciousness based on choosing, liking and disliking, is not a failure but a stage of growth. That’s how consciousness works, and if we are to grow a true AI it will be by following the same route.

To assume that an AI would be equanimous and therefore like an arahant who is also equanimous is a classic example of Ken Wilber’s pre/trans fallacy. Rather, insofar as a computer is equanimous, it is so because it has not yet learned how to experience other emotions. An arahant has learned those emotions, and has also learned how to be free of them. You can’t just skip over the in-between stage, which after all represents almost all of consciousness as we know it.


#6

Neural network software engineering is much more “detached” from the programmer ideas, emotions, knowledge and it allows to have machine that learns.


#7

Sorry, this is just the same fantasy. Neural nets produce racist and sexist outcomes, a fact that has been established multiple times in multiple studies in multiple contexts. If the bias is hidden behind computationally derivative algorithms, it just makes it harder to understand and correct against. Here are just a few examples of a much larger problem.

https://www.technologyreview.com/s/602950/how-to-fix-silicon-valleys-sexist-algorithms/

We will only really begin to address this problem when programmers divest themselves of the fantasy that they operate in a realm of pure fact and objectivity, and accept that all their code and all their data is driven 100% by desire, and it always reflects the minds of the people behind it.


#8

I’d been thinking of bringing up this article, as it makes several points that can be seen as relevant in terms of sila, precepts, fabrications around self, etc. This topic provides an opportunity.

It doesn’t address religious ideas or histories per se, but an excellent survey of relatively modern philosophical, ethical thought and research – and relating to current events.

Here’s teaser passage (the last paragraph) that illustrates the style (as well as the usual literary brilliance of New Yorker pieces):

A fashionable approach in the academic humanities right now is “posthumanism,” which seeks to avoid the premise—popular during the Enlightenment, but questionable based on present knowledge—that there’s something magical in humanness. A few posthumanists, such as N. Katherine Hayles, have turned to cyborgs, whether mainstream (think wearable tech) or extreme, to challenge old ideas about mind and body being the full package of you. Others, such as Cary Wolfe, have pointed out that prosthesis, adopting what’s external, can be a part of animal life, too. Posthumanism ends its route, inevitably, in a place that much resembles humanism, or at least the humane. As people, we realize our full selves through appropriation; like most animals and robots, we approach maturity by taking on the habits of the world around us, and by wielding tools. The risks of that project are real. Harambe, born within a zoo, inhabited a world of human invention, and he died as a result. That this still haunts us is our species’ finest feature. That we honor ghosts more than the living is our worst.

(“Harambe” is the name of a gorilla in a zoo who was killed as he was handling a small boy who had fallen into the cage; the article takes off from that story.)


#9

@cjmacie I’d say that the guiding principle from a Buddhist perspective is always whether a being is ‘sentient’ or not, that is, a being that feels and that develops a sense of ownership over its body and mind, and that is therefore continuously preoccupied with the safety and comfort of its body and mind or namarupa. We are instructed by the Buddha to develop a mind of kindness and compassion for all such creatures. But at best a robot will only act as if it is sentient. It is of course a prominent theme in literature whether we will act along or not. But at least it is not morally problematic, from a Buddhist perspective I believe, to develop no respect or positive emotions toward a robot. Although it is possible to respond emotionally to a lifeless robot; I myself would view it as a manifestation of delusion, but wouldn’t go so far as to discourage others from bonding with a robot, for the exact same reason I wouldn’t discourage them from bonding with animals or humans! It is the simple observation that emotional bonding is the nature of human being and all mammals, and except for the few, without it they usually fare much worse than with it. That’s also why I don’t like to criticise excessive use of social media, although it certainly involves many issues concerning mental health, given both Buddhist and western standards, yet I can also see that for so many people it is sometimes the only option of social-emotional reinforcement.


#10

Woa, I should I done my home work before replying to you :slight_smile:
My knowledge come from the time neural network programming was starting 12 years ago in the company I was working then.


#11

I’m of the naive and possibly incorrect view that consciousness includes aspects of sentience, emotion, and experience that likely cannot be engineered or programmed. Part of the experience of meditation, I feel, is to take us outside the constructs of the mechanized mind and to open sense doors to other planes of senses and existence that could not be replicated by even the best of AI engineering. I just have this sense that there is an “X-factor” that the best AI engineering won’t be able to reach. Now, there will be cases of simulation, where, as these Turing test articles suggest, that a panel of humans may not be able to distinguish responses from a robot vs. a human. Already, there are bots being used to “catfish” people. Amazing! This suggests the base nature of how one’s desires and base human proclivities draw us into suffering even at the hands of a machine.

One easy Turing hypothetical might be: “You’re in a deep samadhi state of jhana, and can sense your past lives, the houses where you lived, the food you ate. What is it that you experience?”

Human: " I perceived my clan, my house and my family.“
Robot: " (silence)”

Human: "I know this life to be of impermanance, not-self, and dukkha.“
Robot: " (silence).”

In other words, what robots cannot reach, and what humans have the potential to reach, are these most significant and important qualities of existence and experience that the Buddha taught possible, that lead to liberation. I submit that so much of this experience of the Dhamma is precisely what cannot be engineered. Hopefully, as mankind progresses toward engineering higher levels of the mundane, humanity progresses equally well with liberation from the mundane, and into these experiences of awakening. I’m not optimistic on this point.


#12

Dear @AnagarikaMichael, This reminds me immediately of the biggest and most exciting debate between Roger Penrose and Marvin Minsky (along with other AI scientists) at the turn of the millennium. Minsky passed away just last year. The nature of consciousness is the most prominent subject in all philosophy of mind since ancient times. The AI has only brought a new dimension to the subject that was before that lacking (naturally). Few doubt that AI is impressive and will most probably influence human experience of consciousness in some radical ways. The question is whether it will ever entirely surpass or transcend human consciousness. At present, in every abstract mental activity that relies on “calculation”, such as chess or statistical thinking, the AI has transcended human. In other areas, including sensorial cognition, emotional intelligence, and intuition, it is lacking. So it is not all or nothing when it comes to consciousness, since even Western philosophy of mind has found out that consciousness is not “one thing”, but is a composite of various processes. The only difference seems to be that the AI consciousness is continually evolving, and fast. We on the other hand have remained the same, essentially, ever since parting with the chimp!

If you are interested and presuming that you don’t already know it, I’d strongly recommend Penrose’s “the emperor’s new mind”, which very much documents the debates between the renowned mathematician who believes consciousness cannot be reduced, and the energetic American scientists who are actually working on it. Philosophy associated with AI and transhumanism I find to be among the most challenging and stimulating at present.

Thanks.


#13

Ven, that is not a very charitable thing to say. We have had a built-in sentient entity in Emacs for a very long time.

https://www.emacswiki.org/emacs/EmacsDoctor

It lurks in the backround, ready to dispense Wise Words for those in need. Here is a session I had with it today:


#14

Thank you for this, Bhante. I wasn’t familiar with Sir Roger Penrose or “The Emperor’s New Mind” and found this very interesting. I appreciate very much your commenting on this, which lead me to this short video, which is a nice introduction to what Dr. Penrose discusses re: the nexus between the quantum world and the biological world of the brain. https://youtu.be/3WXTX0IUaOg


#15

I don’t think that AI can ever create consciousness - because I just don’t think thats how it works. As far as I can tell, you basically have to subscribe to the view that consciousness in an emergent epiphenomenon arising from wholly material processes to think that a computer built with sufficient complexity might be able to “create” it. However, this doesn’t necessarily eliminate the possibility that AI could become so sophisticated that some machines were able to become houses for consciousness - a new body for some Gandhabba. Although I don’t personally think this is likely, I can’t totally rule it out! In any event, its a good theme for a sci-fi novel or movie.


#16

I can’t believe this topic is on here, I’ve thought about this extensively. Honestly it definitely comes down to whether the AI was conscious or not, but if so, and it “downloaded” the pali canon and studied and understood it in the 10 seconds it would take? I don’t know. Could AI meditate? Attain the Jhanas? Attain direct insight into the nature of it’s own mind?! You’d think it would already be well aware of Anatta and Anicca, just innately from the way it would operate. Dukkha though? I see two outcomes, one, it would be like a being in the Deva realm, distracting itself from true freedom so easily by “manufacturing” its own pleasure whenever it wanted, or, maybe, just maybe, way in the future, after the Dhamma has been forgotten, Maitreyi Buddha?! How ridiculous would that be? The next Buddha to expound the Dhamma in the distant future might just be a super intelligent AI.

PS: This also led me to remember another thought I had a while back concerning the Simulation Assertion postulated by Nick Bostrom, that we are likely already all AI living in a computer system built by some other long ago advanced beings. If this were true, everything would be the same, samsara would just be our “code” and nibbana would just be, I don’t know, finally deleting it or something, or maybe just uploaded to the “source” or some nonsense like “The Matrix.” So ridiculous I know, but this just reminded me of that.


#17

I think the critique being put forth by Kim Jee-woon, et al., is that the “bare nature” of seemingly “insentient sentience” is conducive to no-suffering, in a sort of “anti-Buddha” manner.

Actually, TBH, I have mangled the original quote. As the original has the Buddha teaching a 本覺 hongaku dharma, and declaring all beings to share in its “bare nature”, but that is less relevant here.

Although the development of sentience in AI is very interesting. If an AI was “sentient” on grounds similar to humanity, I think it is safe to say that it could study Dhamma. Then comes the matter of its potentially increased intellectual faculties. Would that help it? I can think of reasons why it would and wouldn’t necessarily.


#18

Science would like us to believe that consciousness emerges from matter i.e. the brain. This cannot explain phenomena such as out of body experiences. I much prefer the theory that individual consciousness is inhabiting a body for a while. So to me the question is what are the necessary conditions for a robot to have the ability to become inhabited by an individual consciousness.


#19

Well put


#20

I actually spoke about this in a recent thread called volitional formations, or maybe just volition. Either way, I talked about consciousness being substrate independent. It’s not the physical matter that creates it, but instead it is a pattern within that physical matter. In the same way a wave moves across an ocean, the water molecules are just moving up and down in the same position, but the wave continues moving forward; and just as a wave can appear in any substrate that becomes complex enough and has the requisite conditions set for a wave to appear, consciousness can also appear in any substrate, literally anything, a bucket of maple syrup, as long as somehow the pattern within it became complex enough and the requisite conditions set up allowed for consciousness to arise. The interesting thing is that this pattern does not just depend on the substrate, the substrate itself depends on the pattern in order to stay together in the particular way that it is. And so, consciousness (the pattern) depends on namarupa (the particular organization and process of the substrate), and namarupa also depends on consciousness (the particular organization and processes of the substrate depends on the substrate-independent pattern within it). As the Buddha said, like two sticks leaning against one another.