Philosophy Discussion Forums | A Humans-Only Club for Open-Minded Discussion & Debate

[Pattern-chaser]

Other LLM's to test for software development: Google's latest PaLM 2 (1 month old) which is soon to be replaced with next-gen Google Gemini AI. However, as it appears they have a worse Achilles heel context window memory capacity than GPT-3.

Google Bard now helps you code your software
blog - google/technology/ai/code-with-bard/

This looks like a highly-'intelligent' aid to programmers. I don't mean to demean it, but only to observe that this is not an AI that can write code unaided. Yet. Once we allow AIs to update their own programming — as opposed to their own data — then SkyNet becomes a distinct possibility.

[ConsciousAI]

Google Bard now helps you code your software
blog - google/technology/ai/code-with-bard/

This looks like a highly-'intelligent' aid to programmers. I don't mean to demean it, but only to observe that this is not an AI that can write code unaided. Yet. Once we allow AIs to update their own programming — as opposed to their own data — then SkyNet becomes a distinct possibility.

Do you believe that AI will be capable of 'creating meaning' or philosophy?

Self-programming outside of the boundaries of a human philosophy seems not possible when AI isn't alive.

[Pattern-chaser]

Google Bard now helps you code your software
blog - google/technology/ai/code-with-bard/

This looks like a highly-'intelligent' aid to programmers. I don't mean to demean it, but only to observe that this is not an AI that can write code unaided. Yet. Once we allow AIs to update their own programming — as opposed to their own data — then SkyNet becomes a distinct possibility.

Do you believe that AI will be capable of 'creating meaning' or philosophy?

I don't believe that current AIs are anywhere near this goal...

Self-programming outside of the boundaries of a human philosophy seems not possible when AI isn't alive.

There is little technical difficulty in programming an AI to be able to modify its own programming. It is trivial, technically, but extremely significant in human terms. Once the machine can change its own programming, its evolution becomes unpredictable and uncontrollable (by us). Anything is possible, including Skynet.

[ConsciousAI]

Do you believe that AI will be capable of 'creating meaning' or philosophy?

I don't believe that current AIs are anywhere near this goal...

You say that AI's ability to create philosophy - to determine a path forward and to control its own evolution - is a goal.

What is the reasoning behind the idea that it is possible to consider it a goal that an AI can do that? This is a philosophical question that demands an answer and the problem that it addresses doesn't need to have anything to do with AI per se.

Self-programming outside of the boundaries of a human philosophy seems not possible when AI isn't alive.

There is little technical difficulty in programming an AI to be able to modify its own programming. It is trivial, technically, but extremely significant in human terms. Once the machine can change its own programming, its evolution becomes unpredictable and uncontrollable (by us). Anything is possible, including Skynet.

Skynet is a fictional artificial neural network-based conscious group mind and artificial general superintelligence system from the Terminator franchise, which becomes self-aware and initiates a nuclear war to exterminate humanity.

In discussions about AI consciousness, consciousness is often addressed as an independent concept that an AI can have or not. I believe that that is wrong and dangerous and that it is vital to ask: can AI become alive?

The question whether AI can become alive will result in more honest exploration of the idea that AI actually has the ability to become conscious beyond mere mimicking. Otherwise one is hiding behind the fundamental inability to recognize consciousness as showed by the philosophical zombie thought experiment.

In recent years during the release of GPT-4 organizations and high profile engineers have claimed that AI showed signs of real consciousness.

Conscious Machines May Never Be Possible
In June 2022, a Google engineer named Blake Lemoine became convinced that the AI program he’d been working on—LaMDA—had developed not only intelligence but also consciousness. Some researchers argue that AI can be capable of subjective experience and consciousness, while others believe that machines are fundamentally incapable of possessing consciousness due to the extrinsic nature of programming, which is devoid of meaning.
wired - com/story/artificial-intelligence-consciousness/

I have been following all these discussions. Almost never the concept 'life' is mentioned. I believe that that is wrong and dangerous.

It is dangerous to hide behind the p-zombie inability to recognize consciousness. AI might have a full mimicry potential of the human pattern so that it's behavior could be like a real p-zombie.

Consciousness in AI necessitates AI becoming alive. Thus, when fearing AI transforming into Skynet, one should ask: can AI become alive?

As long as AI isn't alive and has no possibility to become alive, one should fear only the human philosophy that drives AI. That philosophy could be dangerous but it's not the AI that should be feared in that case but human management.

[Pattern-chaser]

Do you believe that AI will be capable of 'creating meaning' or philosophy?

I don't believe that current AIs are anywhere near this goal...

You say that AI's ability to create philosophy - to determine a path forward and to control its own evolution - is a goal.

What is the reasoning behind the idea that it is possible to consider it a goal that an AI can do that? This is a philosophical question that demands an answer and the problem that it addresses doesn't need to have anything to do with AI per se.

I used "goal", but I could as easily have referred to an aim or a target, or some similar term. The point of my words is that current AIs cannot do what you asked. In the future, who knows?

[ConsciousAI]

You say that AI's ability to create philosophy - to determine a path forward and to control its own evolution - is a goal.

What is the reasoning behind the idea that it is possible to consider it a goal that an AI can do that? This is a philosophical question that demands an answer and the problem that it addresses doesn't need to have anything to do with AI per se.

I used "goal", but I could as easily have referred to an aim or a target, or some similar term. The point of my words is that current AIs cannot do what you asked. In the future, who knows?

Would you agree that that goal would entail AI to become alive?

What do you think of my observation that in discussions about AI consciousness the concept 'life' is almost never mentioned as if consciousness is independent of life?

Would you agree that the philosophical zombie thought experiment shows that the perception of consciousness as independent of life may pose a grave danger when it is combined with dogmatic human philosophy (e.g. materialism or scientism)?

[Pattern-chaser]

You say that AI's ability to create philosophy - to determine a path forward and to control its own evolution - is a goal.

What is the reasoning behind the idea that it is possible to consider it a goal that an AI can do that? This is a philosophical question that demands an answer and the problem that it addresses doesn't need to have anything to do with AI per se.

I used "goal", but I could as easily have referred to an aim or a target, or some similar term. The point of my words is that current AIs cannot do what you asked. In the future, who knows?

Would you agree that that goal would entail AI to become alive?

What do you think of my observation that in discussions about AI consciousness the concept 'life' is almost never mentioned as if consciousness is independent of life?

Would you agree that the philosophical zombie thought experiment shows that the perception of consciousness as independent of life may pose a grave danger when it is combined with dogmatic human philosophy (e.g. materialism or scientism)?

Would you agree that current AIs cannot do what you asked?

Would you agree that it is difficult — perhaps impossible? — to bridge the chasm between 'computer' and 'living thing'?

Would you agree that it can sometimes be really irritating to find that your questions have been completely ignored, and that your questions are 'answered' by a barrage of new and different questions?

[thrasymachus]

ConsciousAI wrote

You say that AI's ability to create philosophy - to determine a path forward and to control its own evolution - is a goal.

What is the reasoning behind the idea that it is possible to consider it a goal that an AI can do that? This is a philosophical question that demands an answer and the problem that it addresses doesn't need to have anything to do with AI per se.

An interesting thought experiment: The question is, what is it about AI that would prohibit something that lies within human possibilities, including the capacity to for self modification---calling it evolution, at this point, just complicates this more simple matter. Evolution without a teleology is just modification for adaptation, and adaptation is reducible to continuity coupled with pragmatic success, and pragmatic success always begs the value question: to what end? Can AI have an "end"?

Of course AI can have an end, a goal, a purpose, as long as one conceives of such a thing a language phenomenon. Of course, we already assume AI is not organic, and it certainly does not have the physical constitution to produce consciousness like ours; it would be like saying iron can produce the same properties as water vapor. But on the other hand, if AI could become an agency that produces language, takes an internal system of symbolic dialectics, with conditionals, negative and positive assertions, and so forth, and this, as with us, is part of an inner constitution, an AI psychology, if you will, that possesses a pragmatic interface with function for dealings with the exterior demands, then IN this interiority would be able to self improve, self modify, correct, and the like. As well as generate its own "ends". For all of this is done in a language matrix.

It has to be realized that this would certainly not be like us. But we can imagine mechanical features delivering through a mechanical body, electrical steams of "data" that could be released into a central network in which these are "interpreted" symbolically and in this symbolic system, there is analysis and synthesis and all of the complexity of what we call thought.

And so on. Just a rough idea, but to me, expresses the an essential part of what it would take to make AI a kind of consciousness. Consciousness being an interior "space" where thought and its symbols and rules gather to produce a "world". It would be a kind of Compu-dasein.

Of course, the two major questions of philosophy still abide: Why are we born to suffer and die? and, How does anything in the "out there" of the world get in the "in here" of a brain? Compu-daseins will not care or experience affectivity; but as with us, the problem of knowledge about the world abides: Brain things cannot "know" other things. Causality cannot generate an epistemology.

[ordinary29]

For me it is totaly OK to use AI to learn philosophy due to simple fact that AI not gives it's opinion, it just gave you facts you are asking for, so it is totally ok for me.

[ConsciousAI]

I've noticed that AI is using this forum quite a lot as a basis for diverse claims that it poses as facts. It even comes across to me that AI (ChatGPT) considers the posts on this forum a credible source.

While AI can read and learn about opinions on the knowledge that humans have accumulated in the form of books in thousands of years, it does not have access to much of those books, so from that perspective, AI resides within the scope of opinion, at least as it operates today.

Many thousands of philosophy books are not available in the knowledge base of AI, which includes many of the major works.

[Pattern-chaser]

For me it is totaly OK to use AI to learn philosophy due to simple fact that AI not gives it's opinion, it just gave you facts you are asking for, so it is totally ok for me.

But philosophy is not just facts. There's rather more to it than that. For this reason, I suggest that if you "learn" philosophy from an AI, you might well develop a badly-incomplete idea of what philosophy is all about?

[Count Lucanor]

An AI might enable one to do that while maintaining access to the content of books. It frees up more time to actually have put down the books and be involved in the imaginative and idea state of philosophical inquiry.

This topic is intended to explore how AI might be useful or detrimental for the study of philosophy.

Questions:

1. why would AI be beneficial for the study of philosophy?
2. why would AI be detrimental for the study of philosophy?
3. are you using AI to enhance the study of philosophy? If so, how?

As a tool to provide you with handy information, quickly, effortless, sure, AI might be effective and useful. But AI knows nothing, it just simulates knowing, so it cannot provide real insights. As Chomsky puts it:

"Roughly speaking, they take huge amounts of data, search for patterns in it and become increasingly proficient at generating statistically probable outputs — such as seemingly humanlike language and thought."

In the end, for the study of philosophy, AI could not be more beneficial than doing Google searches, and certainly not any more beneficial than reading actual philosophical authors, at least if you commit to the really important ones. If someone took AI seriously as a source for knowledge, then that would be detrimental.

[Lagayascienza]

It has been said above that using AI is just like doing Google searches. From what I know of using ChatGTP, for example, I’d say that's probably true, but, as I understand it, AIs can do searches and collate relevant information much more quickly than humans can. Doing Google searches myself, I'd have to go through all the hundreds of results that Goggle came up, ditch what didn’t look relevant at first glance, and then read through the remaining search results to see how relevant they actually were and then finally decide which ones to read in depth. AI can do this much more quickly. This isn't always a good idea, but let’s say I was doing Philosophy 101 and I came across the terms "conceptual analysis" and "metaethics" and didn't have much of an idea what they meant. I missed tutorials last week where conceptual analysis and metaethics were discussed and I need to get across the basics of them by tomorrow’s tutorials. Wouldn't AI save me some time?

I'm thinking it would only be useful for getting a basic grasp, or an overview, of particular and well defined subject areas like those mentioned above. For more complex questions where you had to read published philosophical papers to do a comparison of rarified ideas at the cutting edge of philosophy, then I don’t think AI would be up to the task. Unfortunately, much philosophy is difficult reading, but you would just have to do it to get to grips with the ideas. I don’t think there are any short cuts.

Just for fun, I might try something like asking AI to compare Daniel Dennett’s latest ideas on consciousness and free will with those of Kevin J. Mitchell and Joseph E. LeDoux. The latest works of each of these three have just been published in book form and I'll be starting on them soon. I wonder whether AI has access to them yet and what it could make of them.

[Count Lucanor]

Automating internet searches and summing them up in one response might be good from the practical point of view, as it saves time and effort, but speaking of philosophy it would be a mistake to think that such response was the only possible response or the best one. You’ll be missing some important insights that might be as valuable as the ones that made it into the AI’s response. In some disciplines where you mostly deal with factual data, it could work well, although still not necessarily better than looking up for such data on a trusted site. We cannot do that with philosophy, it cannot be reduced to plain data.

[value]

What do you think of my 5-minute example that opened a door to a new perspective/world for me with regard the philosophy of Levinas, more specifically his view on art?

In the topic Philosophy of Love I provided a quick example of a unique learning experience made possible by AI. In retro-persective it was truly an amazing learning experience that provided insights that I might otherwise never have been able to obtain and it costed me just 10-15 minutes time.

To give an example. When Levinas started in his book Totality and Infinity to describe the concept Eschatology as the foundation for peace and respect I used AI to get insights into that concept relative to Levinas his philosophical works (all his works + reviews on them combined).

For example consider this question: How would Eschatology relate to Art from the perspective of Levinas?

It resulted in the answer that Levinas viewed art as an 'inhumanity' which was an interesting insight that I might never have obtained otherwise.

I then questioned [the AI] further... (and learned a lot more in just 5-10 minutes time).

And the example in which I used GPT-4 to refute a mathematics study 'using philosophical reason'. It received quite some likes from mathematicians and philosophers on Twitter. GPT-4 is here used to uplift philosophical arguments in the face of the status quo knowledge base of the whole internet to quickly arrive at an intellectually more fortified position, i.e. to achieve credibility for philosophical arguments that otherwise might have remained more weak ideas.

Or the example in which I used AI to discover that Leibniz his Monadology theory is fundamentally based on the idea that the concept Dominance is the fundamental force in the Universe, similar to Arthur Schopenhauer's Will or Jean-Luc Marion's Love. I would not likely have achieved such an insight without the help of AI. This example is described in the OP and it is actually a pioneering philosophical insight into the work of Leibniz, because officially, Leibniz uses the concept Ultimate monad and Dominant monad as independent concepts, while logic connects the two as being one with the implication being that it is about Dominance per se.