Last Saturday, a good friend, Alberto Ray, and I were having a late night talk. Our talk was very unstructured and revolved on a diversity of non related subjects. So, suddenly he told me that he was convinced that history advances, not necessarily according to a theory of punctuated equilibria reminding us of a Kuhnian process of paradigm’s replacement (he was aware of my passion for Kuhnian-style explanations of change). No, nothing like this. Such a mechanism (or any other similar one) would only be a convincing explanation during the early moments of a technological revolution. But once the buzz caused by the arrival of the new technological (or techno-economic, to use Carlota Perez’s terms) paradigm ends, history would progress in a linear mode. However, from time to time, those linear paths of progress would swerve. Something, at times the least unexpected factor, would make them swerve. Slightly changing the direction of what otherwise would have been a predictable straight path. Regarding the cause of those swerves. He did not make any further explanation. It was then that I told him that his idea reminded me of the concept of clinamen. I vaguely remembered that concept from my reading of Lucretius, The Rerum Natura. When he had coined this concept, originally, his intention had been to defend Epicurus’ atomistic conception of the world, that to him was made of atoms collisioning in the void as in the modern model of the Brownian movement. Since such a movement seemed to Lucretius to be too deterministic, he added the swerve, the clinamen. This concept would make the world more interesting and quite less unpredictable. But last Saturday, I did not have the concept fresh in my mind. So, I interrupted the talk and asked Bard about the definition of clinamen.
LD: Hi Bard, could you please define clinamen?
Clinamen is a Latin word that means «swerve» or «inclination.» In the philosophy of Epicurus, clinamen is the random deviation of atoms from their otherwise deterministic path through the void. This swerve is necessary to account for the existence of free will and for the possibility of novelty in the world. Without clinamen, all events would be predetermined, and there would be no room for human choice or for the emergence of new things.
The clinamen is also important in Epicurean physics. It is the clinamen that allows atoms to collide and interact with each other, which is necessary for the formation of complex objects such as planets, stars, and living beings. Without clinamen, the universe would be a static and unchanging place.
The clinamen is a controversial concept, and there is no consensus among scholars about its precise nature or its role in Epicurean philosophy. However, it is a fascinating and important idea that has had a significant impact on Western thought.
Here are some additional things to know about clinamen:
• It was first proposed by the Greek philosopher Epicurus (341-270 BCE). This is not true. Actually, as I said above, it was Lucretius (c. 99 – c. 55 BC), who was born nearly 250 years later, who coined the term. He wrote in his quoted book:
«Herein I would fain that you should learn this, too, that when first-bodies are being carried downwards straight through the void by their own weight, at times quite undetermined, and at undetermined spots they push a little from their path: yet only just so much as you could call a change of trend. But if they were not used to swerve, all things would fall downwards through the deep void like drops of rain, nor could collision come to be, nor a blow brought to pass for the first-beginnings: so nature would never have brought aught to being.»Lucretius, On the Nature of Things, translated by Cyril Bailey, 1910, London: Oxford University Press). It is of an extreme beauty the way Lucretius expresses his idea: «at times quite undetermined, and at undetermined spots, they push a little from their path». This double underdetermination, certainly makes the world more unpredictable and interesting.
• It is often contrasted with determinism, the view that all events are predetermined.
• It has been interpreted in a variety of ways, including as a physical phenomenon, a metaphysical principle, and a metaphor for human freedom.
• It has been influential in the development of chaos theory, the study of complex systems that exhibit unpredictable behavior.
I hope this helps! Let me know if you have any other questions.
As readers can realize, Bard was only partially correct. However, some of the statements it added around the concept had the ability to trigger further thoughts. Since the ideas were inviting, I made Bard another question:
LD: Do you think that in the inner workings of a large language model (LLM), could there be events analogous to a clinamen?
Bard: Yes it is possible that there could be events in the inner workings of a large language model (LLM) that are analogous to the clinamen. For example, the LLM might be trained on a dataset that contains some noise or errors. This noise could cause the LLM to make unpredictable choices, such as generating a word that is not in the dataset or following a different path through the training data.
Another possibility is that the LLM might be using a stochastic algorithm, such as a random forest or a neural network. These algorithms are designed to make decisions based on a probability distribution, which means that they are not always deterministic. This means that the LLM’s output could be influenced by random factors, which could make it unpredictable.
Talking of non determinism (it came to my mind that nice character, Laplace’s demon (1), and its impossible task), I decided to ask Bard another question, slightly related to the previous subject.
LD: Do you think it would be possible that some of (by hypothesis) the non deterministic events in a large language model could be interpreted as sparks of free will?
Bard: That’s an interesting question. It’s possible that some of the non-deterministic events in a large language model could be interpreted as sparks of free will. However, it’s important to remember that free will is a complex concept, and there is no one definition that everyone agrees on.
Some people might argue that free will requires the ability to make truly unpredictable choices. If this is the case, then it’s unlikely that there would be any sparks of free will in a large language model, as these models are deterministic systems.
However, other people might argue that free will simply requires the ability to make choices that are not completely determined by our environment or our genes. If this is the case, then it’s possible that there could be some sparks of free will in a large language model, as these models are capable of making some non-deterministic choices.
Ultimately, whether or not there are sparks of free will in large language models is a matter of philosophical debate. There is no easy answer, and it’s likely that the answer will depend on how you define free will.
As for Bard, the experiment, I think it’s still too early to say whether or not there are sparks of free will in me. I’m still under development, and I’m learning new things all the time. It’s possible that I’ll develop the ability to make truly unpredictable choices in the future. But for now, I’m content to be a deterministic system that is capable of generating creative and informative text.
LD: Thank you Bard, nice talk.
Bard: Thank you! I’m glad you find my answers helpful. I’m always working to improve my ability to generate creative and informative text, and I appreciate your feedback.
I’m also interested in the philosophical implications of large language models, and I think it’s an important topic to explore. I’m not sure whether or not there are sparks of free will in large language models, but I think it’s a fascinating question to ask.
Thank you for your question!
I do hope Bard’s answers were also stimulant to you. Non Determinism and free will have always been subjects deserving of my interest. Although, it seems obvious that neither Bard nor other LLMs have developed free will, the short talk made me wonder whether there were, deep woven inside some of the most advanced LLMs, sparks of free will. If that would be the case, I wonder whether those fleeting moments of free will could ever become a permanent trait of existing or future LLMs. Being that the case, a mechanism should be developed to inform the owners /developers of the AI, thatthe system is showing evidence of that trait and make the decision regarding its eventual consequences.
Specifically regarding the talk I had with Bard, I would only like to emphasize that the term clinamen was not intended to be used as an explanatory concept of the mechanisms of history my friend was attempting to describe. It just happens that my mind decided to make a whimsical shift of subjects, from History and the rules that determine its path, to the nature of reality and, from that, to the potential of free will within an AI. Meanderous digression.
(1): Here the quote from Pierre Simon Laplace, A Philosophical Essay on Probabilities (Truscott, F.W. and Emory, F.L., Dover Publications, New York, 1951)
We may regard the present state of the universe as the effect of its past and the cause of its future. An intellect which at a certain moment would know all forces that set nature in motion, and all positions of all items of which nature is composed, if this intellect were also vast enough to submit these data to analysis, it would embrace in a single formula the movements of the greatest bodies of the universe and those of the tiniest atom; for such an intellect nothing would be uncertain and the future just like the past would be present before its eyes.
What Laplace called as an intellect, the philosphers called a demon.
caracas 10N, 67W 