Philosophy and the question 'When is a robot considered alive and thinking?'

Question

I originally posted this question in Robotics Stack Exchange, but it was deemed off-topic there and it was recommended I post here:

I was watching a youtube video of a robot:

https://www.youtube.com/watch?v=rVlhMGQgDkY

At a point in the video when the man pushes the robot, I felt sorry for the robot. It made me think, at what point do we class a robot as alive? Without going too off-topic; if you're an atheist and you think that 'life' is merely the chance result of chemicals under changing conditions - then what makes a robot different from humans? I assume it's to do with complexity - so is there some level at which we start to see the robots as an entity that should have rights? It seems to me irrefutable that we experience the Uncanny Valley.

Broadly, how do the philosophies of biology and mind address a question such as 'When is a robot considered alive and thinking?'

Answer 1

The definition of life is one of the most famously difficult definitions in all of philosophy. There are many definitions. For example, science has a descriptive definition for life:

1. Homeostasis: regulation of the internal environment to maintain a constant state; for example, sweating to reduce temperature
2. Organization: being structurally composed of one or more cells — the basic units of life
3. Metabolism: transformation of energy by converting chemicals and energy into cellular components (anabolism) and decomposing organic matter (catabolism). Living things require energy to maintain internal organization (homeostasis) and to produce the other phenomena associated with life.
4. Growth: maintenance of a higher rate of anabolism than catabolism. A growing organism increases in size in all of its parts, rather than simply accumulating matter.
5. Adaptation: the ability to change over time in response to the environment. This ability is fundamental to the process of evolution and is determined by the organism's heredity, diet, and external factors.
6. Response to stimuli: a response can take many forms, from the contraction of a unicellular organism to external chemicals, to complex reactions involving all the senses of multicellular organisms. A response is often expressed by motion; for example, the leaves of a plant turning toward the sun (phototropism), and chemotaxis.
7. Reproduction: the ability to produce new individual organisms, either asexually from a single parent organism or sexually from two parent organisms.

Of course, there is no particular rule which says everyone abides by this definition.

An interesting subjective answer would be that a robot is "alive" when you believe it is alive. This resolves a remarkable number of questions, but leaves the pesky subjectivity in place -- whether or not something is alive is a question people can disagree on if this definition is used.

Of course, if the issue is that you felt sorry for the robot, you can always reframe the problem and make sure the robot does not deserve your compassion because it is evil.

Answer 2

There is a distinction between being alive and being sentient and both categories have their own criteria. Modern philosophers, particularly in the analytic and Continental traditions of philosophy rely on these scientific definitions. Whether or not a robot is alive is really not very controversial, because robots as traditionally conceived fail a number of philosophical criteria raised in the philosophy of biology. Robots are machines that don't metabolize or reproduce according to those biologically oriented philosophical standards. The question of sentience is a little more complicated and philosophers who work in the philsophy of mind find answers to be controversial. That is, there is wrangling occurring over what constitutes an adequate theory of mind and consciousness.

The first to offer rigorous criteria was Alan Turing who suggested what is now known as the Turing Test. Some philosophers like Dennett with his eliminative materialism see the mind as an illusion. There are many philosophers of mind who are recognized as authorities, such as Margaret Boden, Paul and Patricia Churchland, and John Searle. One of the most famous philosophers regarding these issues is David Chalmers who proposed what is known as the Hard Problem of Consciousness.

It is fair to say that the question of sentience applies to animals as much as to robots. There is a reasonable consensus among biologists that an organism is sentient if it fulfils two conditions:

• Its brain contains the necessary organized physical structures to support sentience, such as a suitably advanced cerebral cortex.
• It exhibits cognitive behaviours such as ones which can only be explained by assuming it has a theory of mind and can envisage what another creature might perceive.

There is no known scientific or strictly logical basis for denying that an artificial mind would be subject to the same criteria, judged against equivalent evidence. In fact, there are very strong philosophical arguments that robots can in principle be consciously aware and think. See PhilSE: Computers, Artificial Intelligence, and Epistemology

So the question boils down to: what defines cognitive behaviour, and what substrate structures are required to support it? These questions currently remain open.

For example, some butterfly wings have evolved eye spots which mimic the eyes of a large predator, i.e. we might understand them as signs or symbols. In a smaller predator stalking the butterfly, these can trigger an evolutionary response, such as flinching, which gives the butterfly time to escape. Such a response might be reflex and need not involve sentience. In AI a parallel would be sticking a small patch of paper on the road to fool a self-driving car into changing lanes. To recognise the spoof as a spoof and the signs as fakes do appear to require cognition, as one must overtly relate something which is not there to what is being observed.

Another example would be the relatively common practice of a corvid bird or a squirrel deliberately burying food in sight of a rival, guarding it until the rival is distracted, and then unearthing it and hiding it again elsewhere, this time in private. It is hard to account for such duplicity without granting a theory of mind, and hard to describe it as anything other than cognitive. But to what extent do the two go hand-in-hand?

It has been suggested that a mental model of one's own self is a necessary prerequisite to modelling other selves, or possibly vice versa, but all such ideas currently lack clarity.

Answer 3

I believe it could be said to be alive, even if it isn't conscious, i.e. it is a zombie. John Searle discusses part of this in the Chinese Room argument. It is mentioned somewhere in Searle: Philosophy of Mind lectures. In Cambridge Handbook of Artificial Intelligence (ISBN 978-0-521-69191-8) there is a section Artificial Life by Mark A. Bedau with more arguments. I guess your question about whether a robot is alive is really about whether the robot is a strong AI and has achieved consciousness, and thus represent a person. The European and American view on this seems to diverge, see for example Politico: Europe divided over robot ‘personhood’ i.e. moral status. It is not so much that Europe is divided, as Europe takes another approach than the USA, European Parliament: REPORT – with recommendations to the Commission on Civil Law Rules on Robotics.

When a robot has a mind of its own, then you should feel bad about pushing it around. ;)

Answer 4

First, there is no absolute definition of life. Any definition that you might find, might probably be applied to a rock, so a rock is a living entity.

For example, taking the concept from another answer:

• homeostasis: a rock is able to restructure its molecules to avoid breaking when it is submitted to a tension or heat (believe it or not: molecules might even change positions);
• organization: atoms, molecules, particles, etc. in any object tend to keep patterns; when you touch a rock, it doesn't disintegrate, this seems normal, but it shouldn't; you are applying a destructive force, and the rock keep trying to maintain order;
• metabolism: well, anabolism and catabolism are subjective ideas about what we consider living entities (tautological: we define living things by behaviors like this, which at the same time, define what life is). Rocks are anyway able to transform energy: put a pressure on a rock, and it will produce heat;
• growth: a rock in a water course filters water within its microscopical structures, and captures mineral particles which with time become part of the rock. Of course, this might take hundreds of years, but it's just a matter of scale. So, we can say that rocks do grow, they just grow very slow, regarding our perception.
• adaptation: some rocks are able to keep integrity when submitted to large pressure, others, when submitted to high temperatures. Rocks that hold on high temperatures might tend to grow, to enforce their structure along years, incorporating more mineral particles, but only minerals that hold upon large temperatures;
• response to stimuli: the idea of a response is very subjective, because it depends on the notion of causality, which is known to be largely biased (see David Hume or Russell on causality). You can say that there will be a response if the result of a stimuli changes in the presence of the object. Therefore, if you speak to a rock, and measure the auditive answer, in the presence, and in the absence of the rock, the echoed result will be different. Then, the rock is able to respond... to your voice. Anyway, rocks might answer to other types of stimuli: they would get deformed with some pressure (there exist even flexible rocks), and regain its shape after removing the stimuli. Following Newton, if you push a heavy rock, you are doing an action, and the rock will perform a reaction: a force against the direction of your force: there you have an answer to an stimuli; etc.
• reproduction: following the idea of growth, a growing rock in a course of water might break. Each one of its parts will keep growing. With time, two new rocks resembling its relative parent will exist.

Second, thinking, it's just calculating. So, if a robot can calculate something that would impact its behavior, it would be thinking.

Therefore, it can be simply concluded that a robot is a living entity and that a robot is a thinking entity.

But you ask for the main difference. That's not how robots differ from any other living entity, like ants. Robots or ants are both thinking and living entities.

The key difference is that robots cannot act with intelligence. That would be the main element of discrimination between both: ants are intelligent entities.

For what this concerns, intelligence is the ability to think and act coherently towards the most important priority and deepest goal: to survive [1][2]. Acting coherently implies that actions are all logically aimed to the main goal. Some people is incoherent, and therefore they do not act intelligently (e.g. some people are aggressive, which implies that they produce rejection on others, the opposite from attraction, and therefore, they tend to be isolated, isolation meaning reducing, even if minimally, the probabilities of survival).

A robot is not able to keep exerting its functions after its battery is exhausted. Yes, some vacuum cleaning robots can reach the electrical plug. That represents a bit of intelligence. But just an error, and they stop existing as such. They depend on us to live. If they suffer any damage, they need of us to get repaired. They can't take decisions towards their survival. They depend on us for that; etc. Perhaps a synthesis of all those concepts: they cannot keep a low entropy by themselves.

Ants, on the contrary, can do all those chores. They can search for energetic supplies (probably their second most important priority). They can take decisions and learn from their mistakes following the goal of survival. They can repair themselves. But most importantly: they keep a low degree of entropy almost by themselves (entropy is not as much as disorder as it is energy dispersal [just check the formulae]; ants can concentrate energy and use it to keep existing).

In conclusion: current robots are living entities and thinking entities. But they are not intelligent entities. The day that robots behave intelligently, they might get quite similar to us in many senses. Perhaps they might get indistinguishable.

[1] What is intelligence?

[2] The concept of intelligence is as difficult as the concept of life. The concept of intelligence used here is not widely accepted, it just seems to fit well to robotics and AI, its part of my experience in philosophy discussion groups (and I also work in computers/AI). The fact that survival is our highest priority is just logical. What decision would prevail to survival? Perhaps dying for the family. But that's just survival of the group, which is equivalent. Suicide doesn't count, because the highest priority in such case is to stop existing, and suicide is the logical choice.

Answer 5

It sounds like your reaction when the robot was pushed over was important in forming your question. Its importance is perhaps due to the activity of your mirror neurons (and other brain events) encouraging a kind of empathy in your mind— since if a robot has a mind it might experience some kind of pain, as you might, when pushed over.

And so, beyond the attempts to define biological life, you seem to be asking a question about consciousness. Such as whether a robot or other computer can have a mind like you and I do? And if so, whether your mirror neurons are detecting a potential mind state (pain) in such a robot.

These questions are part of Philosophy of Mind. https://en.m.wikipedia.org/wiki/Philosophy_of_mind

To begin your own studies on this question it might help to start with a famous thought experiment called the Chinese room, which tries to show that minds are more than the computations or functions they perform. I.e., that a mind is not simply a system of computation or collection of “mechanisms”. https://en.m.wikipedia.org/wiki/Chinese_room https://plato.stanford.edu/entries/chinese-room/#SystRepl

However, because this is a highly debated topic, it may be best to read the various philosophies of mind and develop an informed opinion.

Answer 6

I covered defining life & intelligence here. Drawing a hard line between life & non-life is not tenable, or useful. 'Thinking' is a bit of a weasel word, because it can contain a raft of hidden assumptions about what it implies. I suspect what you have in mind is self-awareness.

The real questions are about capacity to suffer, and moral value considerations. Bostrom addressed issues around unexpected or unknown suffering in AI systems, and the concept of generalising moral considerations beyond biological consciousnesses in his book 'Superintelligence', the latter specifically in his malignant failure mode, mind crime. Marvin The Paranoid Android is a humourous but pertinent illustration of AI suffering: “Here I am, brain the size of a planet, and they ask me to take you to the bridge. Call that job satisfaction, 'cause I don't.”

Embodied Cognition is an important consideration, the intelligence contained in our physical bodies, social structures, and intelligence beyond cognitive systems such as our hormone production, and ecological intelligence/s. These can be simulated too, of course. But they indicate challenges for synthetic intelligence, especially of scale. The idea of 'salience landscapes' and 'cognitive grip' (see Vervake) help understand this, that successful abstraction is about forming cognitive models that suppress noise or irrelevant data, and highlight important things in task-orientated ways. Cognitive grip is in direct analogy to how our hands are useful, and we develop skills with them.

Human language is a kind of distributed collaborative intelligence, which structures and focuses salience landscapes, and supports development of cognitive grip. We are eusocial animals, hive-like. And intersubjectivity, making the leap-of-faith other minds are like our own, and mentally projecting ourselves into the subjectivity of others and inviting them into ours, is key to human communication. Octopuses are highly intelligent yet solitary, so we know sophisticated problem is not solely social, but it is highly correlated. Octopuses are about as structurally different to ours as we can imagine, but interactions with them suggest a deep ability to connect with human minds (see Other Minds). A strong case can be made that any robots or artificial general intelligence 'superintelligences' will emerge in our information ecoligy, and participate in our distributed intelligence (ie, language).

The metaphor of Indra's Net is illustrative of intersubjectivity, and how collective awareness can be magnified through reflection (interaction) of other minds. I have high hopes for the conceptual framework of peer-to-peer reality, which I would summarise as there is no objectivity, only reified (abstract thing made or treated as concrete) intersubjectivity. This tells us something about the likelihood of being able to interact with aliens, and that the idea of drawing a hard line between humans and artificial general intelligences is probably musguided. AGI & human intelligence will develop collaboratively together - that would be true even where collaborating means involvement in conflict (Raised By Wolves series nicely illustrates this).

Answer 7

Then there is the Turing Test. What if you simply could not tell the difference between the robot’s actions and those of a human being? Would the robot be alive for all practical purposes?

Alan Turing proposed The Imitation Game in 1950. Generally, the game addressed the question of whether a computer can think, while avoiding problems that can arise when defining the terms of the question.

Turing proposed that a human evaluator would judge natural language conversations between a human and a machine designed to generate human-like responses. The evaluator would be aware that one of the two partners in conversation is a machine, and all participants would be separated from one another... If the evaluator cannot reliably tell the machine from the human, the machine is said to have passed the test.

Wikipedia, Turing test.

Now expand the test so that the robot operates for as long as a human lifetime. If the observer cannot distinguish the robot from a human, the robot has passed the test.

Answer 8

Robots would get this status if they became viable as everyday companions, as well as similarly irreparable after damage.

By companions I mean at least the degree of companionship of pets like cats and dogs (though it could be much more). By irreparable I mean that damage to their brain or body could not be easily fixed.

Without companionship, robots will remain tools to us. Without vulnerability, ethical discussions are moot because all damage could be completely undone (very differently from biological life).

Answer 9

Robots will never be human beings. The robot, by very definition, is a machine.

"A robot is a machine—especially one programmable by a computer— capable of carrying out a complex series of actions automatically." -- Wikipedia, Robot

Even if you have the most sophisticated android, capable of thinking and acting like a human being, a robot will always be and remain a machine. Life is not a machine.

Answer 10

We all know the main characteristics of a living thing. Can we ever expect all such characteristics in a robot also? IMHO, it would be impossible. Its body would always be made up of non-cells. If it is made up of cells, it would never be a robot. So in its micro-level it has no life at all. Since it has no life in micro-level we can never say, 'this robot is alive'. But since its functioning resembles a living thing we are compelled to think so.

Let us suppose: 'To feel like a living thing someday perhaps scientists would give life to some of its parts.' Even then we can't say it is alive. We can only say that some inner part is alive. We should not forget that this is not the case of living things. Almost all their cells are alive. So we can always say the thing itself is alive.

But the word 'alive' has another meaning -- alert and active. In that sense only we could say that 'this robot is alive' (if it is switched on and is working. No matter whether its functioning wrongly or not).

Even if we install AI on a robot very effectively, it will only respond as programmed. We can't expect any innovation from it. Any way, if it can perceive our feelings even by an entirely different means, we could say that it is sentient. But it will never have life even in a slightest amount. So, if we are compelled to say that it has life, we would also be compelled to introduce a new biology with weird/new fundamentals.

Let us think about its thinking process.

Though the initial process (analysis) for its working is not thinking as we usually refer, my opinion is that we could say that it is thinking. I didn't forget the fact that the thoughts of a living thing affect each of its cell minutely at least. But here, in the case of a robot, thinking does not affect so. So we should say even though it is thinking, the two thinking processes are entirely different.

Even the same thought is seen in robots of the same type, whereas only similar thoughts can be seen in living things at maximum. This definitely shows that a robot's thought never reaches the level of mind or the subtlest form for thought.

Again, we have already named 'the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions' as AI. More over, this term also refers to any machine that exhibits traits associated with a human mind such as learning and problem-solving. So one could argue: "To exhibit intelligence -- the ability to acquire and apply knowledge and skills, thinking and analysis etc., are necessary. So, also in the case of a robot, we are compelled to use that term -- thinking." We are helpless here. Now that we have already accustomed to this different usage in our day to day life, we can say a robot is thinking (can think) without mind (which is only a bundle of thoughts). Even then don't forget that thinking happens in humans even if nothing is preinstalled deliberately. But this never happens in a robot. So its thinking has not all the things needed for a thinking process.

CONCLUSION:

Your question can be taken mainly in two sense.

• 1. (When these two characteristics are taken separately)

A robot can be considered as 'thinking'; but can't be considered as alive as we usually refer.

I mean, even when we use the term -- 'thinking' IN ALMOST ALL CASES of a robot, we can't say 'it is alive' in future also.

• 2. (When these two characteristics are taken together)

A robot can't be thinking since it can never be alive.

=========================================================================

This question has another aspect, but almost all people don't care about that. There, Philosophy of mind 'absorbs' Philosophy of Biology. That situation is mentioned below.

When one realizes what consciousness is, his views about non-living things would change completely. Then the answer to this question would also change. "There is consciousness (also the thing behind thought) everywhere. So there is no difference in living and non-living things.", he would say.

Answer 11

SHORT ANSWER: There are no thinking robots. The robot in that video, however, is controlled by its Neural Network processor, and it is only natural for humans, who are themselves run by neural networks, to empathize with other NNs -- humans, animals or, indeed, the robots.

LONG ANSWER: NN is one of the two centers of cognition in human brain. The other one is the rational Self, responsible for knowledge, the concept of truth, self-awareness (by modeling themselves as a part of a comprehensive reality simulation), and consciousness (which is actually about the agency, as in making choices consciously). It is also very much optional -- most people are being run by the lone NN. But even with the Self in charge, it is too slow to run a human in real time. Rather, it's job (aside from the long term planning and maintaining mental simulation of reality) is to analyse the recent events, evaluating the NN performance, and learning the lessons. To the Self it looks like reflections in solitude, but NN is always listening and taking pointers.

Below we discuss human traits that originating from NN, and well as it's character and epistemological aspects.

A Neural Network: technical perspective.

First, let's clarify what NN is not. It's not your typical von Neumann architecture with clearly defined functional components and designed to execute algorithms:

An neural net does not run algorithms or operations -- it's a self-learning image classification system. It learns by trying to guess which class a particular image belongs to. Rather than storing the image itself, it stores similar patterns/anti-patterns for each class. Then it looks for familiar patterns the encountered image to determine it's class. It's blazing fast and memory-efficient despite its low complexity.

In terms of behaviour, the above would produce the following traits/abilities:

• Intuition. That is, essentially, what neural networks do. They never know anything for sure, so it's always a guesswork based on experience. And since it is not a result of an algorithm, it is unexplainable... like that gut feeling.
• Emotional responses in humans and animals, including empathy.
• Qualia, a.k.a Locke's Simple Ideas, a.k.a. Kant's intuitions, I prefer "concepts" -- collections of patterns/anti-patterns matching a particular class. That's why, for example, an attempt to describe what a "chair" is mostly results frustration, even tho you know what a chair is don't you?.. Well, actually you don't -- but you might have a pretty good idea of the "chair".
• Creativity -- because intuition is guesswork, if NN have no clear idea, it picks a probable outcome, or simply picks at random. Call it "a leap of faith" -- a drunkard searching for the keys under the lamppost is one example. Assuming the existence of objective reality we all share is another.
• The sense of beauty, aesthetics. Its rational equivalent concept is known as "efficiency". I.e. "beautiful" is what NN perceives as "efficient" (and, in some cases, literally).
• Subjectivity -- relying on its personal experience alone makes NN 100% subjective. To NN there is no such thing as "outside", making the objective reality just as incomprehensible.
• Very superficial. Being an image recognition system, NN makes choices based on superficial appearances.
• Irrationality -- it's always guessing and hedging her bets, NN knows nothing and understands nothing.
• Selfishness. NN = unexplainable AI. It cannot explain its own beliefs, much less understand other's perspectives other than its own. Unable to reconcile its perspective with that others leads to power struggle, as the only conflict resolution strategy.

Call her the "beautiful" mind -- I'm sure we all know the type :)

A bit on our Selves

Being humans, we've got an alternative cognitive faculty at our disposal. The Self should be in charge at the executive level, using NN as an autopilot/flight computer (NN only knows how to how to learn and imitate a behaviour, and normally it imitates the Self). Unfortunately, the sad realities of the nuclear family apocalypse we've been living since the original sin of 3,000 years ago, force the Self to quit while still in their childhood, leaving NN autopilot "ON" forever.

That’s why most people end up pure NNs, 100% irrational -- immitatting rationality when they can (i.e. happen to have enough of the hands-on experience, the small things of everyday life); but falling flat on their faces now and again when it comes to the big questions, social policies in particular.

After all, few have a hands on experience running a country or resolving international conflicts.

"Do you know, my son, with what little understanding the world is ruled?"
  -- Axel Oxenstierna, Lord High Chancellor of Sweden from 1612 to 1629.