Tolkien on whether things that can talk have to have a soul

[KnightFluttershy]

As Tolkien was a great expert on linguistics, I have often found myself wondering what his opinions on generative text LLMs like "CHAT-GPT" would be. I will compile all the quotes here that I think are relevent, and then try to end it with my own explanation as to what I think he would have said.

1. Letter #153, September 1954,

(In this letter Tolkien is asked whether the fact that trolls can speak means that they have souls)

Of course (since inevitably my world is highly imperfect even on its own plane nor made wholly coherent – our Real World does not appear to be wholly coherent either; and I am actually not myself convinced that,though in every world on every plane all must ultimately be under the Will of God, even in ours there are not some 'tolerated' sub-creational counterfeits!) when you make Trolls speak you are giving them a power, which in our world (probably) connotes the possession of a 'soul'.

2. A text from Morgoth's Ring, 1959

True 'rational' creatures, 'speaking peoples', are all of human / 'humanoid' form. Only the Valar and Maiar are intelligences that can assume forms of Arda at will. (...) In summary: I think it must be assumed that 'talking' is not necessarily the sign of the possession of a 'rational soul' or fëa

My explanation

I believe that, from what we can garner here, Tolkien ultimately came to the conclusion that talking did not neccessarily indicate the presence of a soul, though he seemed to have been conflicted about this for a few years. Ergo, if he came to the future and heard about our modern day LLMs, I suspect he would not be quite as surprised as many modern people were by them. Perhaps these were the "sub-creational counterfeits" he was referring to!

P.S. This is a personal addendum, but I have never liked the idea that animals don't have souls. C.S. Lewis did a brilliant essay entitled "On Vivisection" that argues for animal welfare, and in "The Problem of Pain" he suggests animal souls could go to heaven and even suggests that a lion might be far more terrifying in heaven than on Earth (this argument in his essay was one of the foundational things that converted me from hardcore atheist -> Christianity). It's not hugely revelvent to the overall point about LLMs (which weren't really a direct consideration of Tolkien's) but I felt I had to include it anyway.

 

[KnightFluttershy]

This post was inspired by a Lord Dunsany story with the quote "And Man said in his heart: "Just one invention more. There is something I want to do with petrol yet, and then I will give it all up and go back to the woods."

 

[Ent]

Ent finds it fascinating no one has, as yet, spoken into this particular post with its magnificent and extensive 'field of vision' open and available for discussion.

 

[Alexander Scamilton]

Ent finds it fascinating no one has, as yet, spoken into this particular post with its magnificent and extensive 'field of vision' open and available for discussion.

Thank you for saying nothing except pointing out how nobody has said anything and preceding to say this is a great topic and then saying nothing on this topic.

As Tolkien was a great expert on linguistics, I have often found myself wondering what his opinions on generative text LLMs like "CHAT-GPT" would be. I will compile all the quotes here that I think are relevent, and then try to end it with my own explanation as to what I think he would have said.

1. Letter #153, September 1954,

(In this letter Tolkien is asked whether the fact that trolls can speak means that they have souls)

Of course (since inevitably my world is highly imperfect even on its own plane nor made wholly coherent – our Real World does not appear to be wholly coherent either; and I am actually not myself convinced that,though in every world on every plane all must ultimately be under the Will of God, even in ours there are not some 'tolerated' sub-creational counterfeits!) when you make Trolls speak you are giving them a power, which in our world (probably) connotes the possession of a 'soul'.

2. A text from Morgoth's Ring, 1959

True 'rational' creatures, 'speaking peoples', are all of human / 'humanoid' form. Only the Valar and Maiar are intelligences that can assume forms of Arda at will. (...) In summary: I think it must be assumed that 'talking' is not necessarily the sign of the possession of a 'rational soul' or fëa

My explanation

I believe that, from what we can garner here, Tolkien ultimately came to the conclusion that talking did not neccessarily indicate the presence of a soul, though he seemed to have been conflicted about this for a few years. Ergo, if he came to the future and heard about our modern day LLMs, I suspect he would not be quite as surprised as many modern people were by them. Perhaps these were the "sub-creational counterfeits" he was referring to!

P.S. This is a personal addendum, but I have never liked the idea that animals don't have souls. C.S. Lewis did a brilliant essay entitled "On Vivisection" that argues for animal welfare, and in "The Problem of Pain" he suggests animal souls could go to heaven and even suggests that a lion might be far more terrifying in heaven than on Earth (this argument in his essay was one of the foundational things that converted me from hardcore atheist -> Christianity). It's not hugely revelvent to the overall point about LLMs (which weren't really a direct consideration of Tolkien's) but I felt I had to include it anyway.

Sure, it doesn't have a soul, but what if it did? What if we gave life to something? Yes, we aren't close to that now with AI, but we will be. Looking at all these stories where man gives life to something (like Ultron or Frankenstein), I think it's bound to happen.

 

[KnightFluttershy]

Thank you for saying nothing except pointing out how nobody has said anything and preceding to say this is a great topic and then saying nothing on this topic.

Sure, it doesn't have a soul, but what if it did? What if we gave life to something? Yes, we aren't close to that now with AI, but we will be. Looking at all these stories where man gives life to something (like Ultron or Frankenstein), I think it's bound to happen.

My favourite example of this is actually from the game Fallout:New Vegas, where you find this robot, called a Securitrom, who at first seems to be able to think for himself, but after you talk to him enough, he just keeps directing you to his creator Mr. House and telling you what a swell guy he is, and you slowly realise that he was never able to think for himself at all and was just a glorified billboard for Mr. House. So they never had the potential to be alive or have souls. The dwarves, when they are created by Aule and don't have a soul, behave in a very similar way, as they can only act as extensions of his will before Eru grants them souls.

 

[Ent]

Thank you for saying nothing except pointing out how nobody has said anything and preceding to say this is a great topic and then saying nothing on this topic.

You are most welcome.
Ent is a master at saying nothing using many words.