Or: me versus ChatGPT.
Or: “I am not an LLM”.
Large language models (LLMs) are great for distilling humans on the Internet into a super spirit of predictability. Given this I’d thought I’d ask ChatGPT:
What questions do human beings most want answered?
Me, asking ChatGPT
Example answers:
Now LLMs are perfect at providing a reasonable but ultimately unsatisfying and clichéd answer to these questions. So I’d thought I’d set myself a challenge: am I better than ChatGPT?
(All images, apart from the first, also generated using StabilityAI’s DreamStudio (Stable Diffusion API)).
1. What is the meaning of life?
Me:
First, note the form of the question. We don’t ask as often: “what does life mean?”
The word “meaning” is used to refer to the significance or purpose that something has. In this sense, “meaning” refers to the ideas, values, or beliefs that are attached to a particular object, symbol, or action. “Meaning” is the semantic coupling we attach to concept of “life”.
When applied to “life” “meaning” is often used to refer to a purpose or significance. We see that a question with a similar aim is: “what is the purpose of life?” By rephrasing in this way, we also see that often the subtext is “my life” as opposed to all life. When we ask what is the meaning of life, we are often implicitly not asking: what is the meaning of the life of a pot plant or fly? This belies an inbuilt natural egotism to our questions in general. The meaning of life is often a deeply personal and subjective concept that each individual must determine for themselves.
In the context of the question “What is the purpose of my life?” the word “purpose” is used to describe the overarching goal or objective that a person aims to fulfill through their actions and experiences. This is again telling. This, plus the original, question indicate we are looking for a singular answer. Douglas Adam’s skewers this in two digits. We want an easy answer – one action we can do and then be swept over by a feeling of peace and satisfaction. When put like this the absurdity is clear.
In actual fact, the regular asking of the question is possibly more important than the answer. We need to ask the question in reference to our own life’s to help structure our mental framework of action. We want a singular reference point to hold as context to constrain all the lower daily decisions we need to make.
ChatGPT:
2. What happens after we die?
Me:
This is easy: nothing.
There’s some heavy lifting being performed by the “we” in the question, with a suggestion of dualism. The “we” that we assign to the flow of thoughts and feelings and bodily sensations ceases to be as our brain dies. The “we” that encompasses our bodies begins to decay. The “we” that refers to our molecular constituents return to the world in some form.
We want to imagine a continuity because our sense of “selfness” is so strong and always present. But this sense of “selfness” is itself built upon sensing of bodily processes that underlie our daily existence.
We also see purpose and agency in inanimate objects. This is because our brains are structured to see social agency everywhere. Our afterlives are faces in the clouds on a summer’s day.
ChatGPT:
3. How did the universe begin?
Me:
We know a fair amount on this one: https://en.m.wikipedia.org/wiki/Big_Bang.
Again, our brains belie us here – we are often looking for a mythical single action that is the casual starting point of everything. It seems disappointing to think that there may be no absolute start, that everything is an infinitely repeating cycle or endlessly extending process. We’d feel especially cheated if two events in this sequence were partitioned informationally, such that we could never know the contents of precursor state.
ChatGPT:
4. Is there a God?
Me:
Definitively, we can be sure of nothing.
To a very high degree of probability (high 90s), I’d say the answer is: “no”, at least to the concept of “God” that 95% of the question askers have in mind.
If you were to define “God” in a Spinozian manner to mean the entirety of everything seen from an external timeless perspective then the answer would be “yes”.
If you were to define “God” as the name of the material manifestation of the collection of concepts relating to “God” that exist, have existed, and will exist in the minds of all believers, then again the answer would be “yes”. This “God” would be “love”, or at least conceptually overlap, and would have a causal influence on believers and those that interact with them, whether that be physically or culturally.
This question also belies a background related to the Abrahamic religions. This is around 40% of the world (3.8B / 10B). To the majority of human beings the capital letter would not make sense.
ChatGPT:
5. What is the purpose of existence?
Me:
This is very similar to the first question. Let’s concentrate on the differences.
Here we are asking about “existence”. Although there is often a silent, implied “my”, let’s take this out and consider “existence” in general. Note here there is no reference to life, so let’s lean into that and consider “existence” as the persistence of any object through time for at least a non-zero period; the object can be animate or inanimate.
To be fair, we’re actually not that bothered about the time aspect of things. We want a singular answer that is independent of time. We don’t really care if we are considering a photo snapshot or thinking of an object lifetime – our brains collapse it all down to a timeless concept anyway.
Just asking the question makes us likely to fall into the traps of the largely discredited philosophical idea of teleology. Teleology is a philosophical perspective that views the natural world as being driven by purpose or design. But this idea gets it inside out – we see project purpose onto objects as our brains are always asking the question: “what can this object do for me?” Indeed, there is a question as to whether objects themselves exist independently of us observing them, not in a “I create the world as magical being” but in that the world (especially the natural world) is full of overlapping things at different scales and so the concept of independent objects entails projection from ourselves.
There is another, more interesting question: why are we so scared of the answer being – “there is no purpose”? Why does this answer leave us cold and searching? Why do ourselves and the objects around us need to fit within a narrative structure reflective of our capacities in action planning?
ChatGPT:
6. Why do we suffer?
Me:
We suffer because we are living beings, cascades of chemicals literally hewn out of rock, fizzing balancing acts that could collapse at any point, blowing our entropic triumphs to the wind.
Pain – whether manifest from clear physical symptoms or more nefarious mental causes – is a signal to direct our attention. It needs to feel “bad” to get our attention and act quickly on potential causes. It also tells us when our body and minds are recovering or I’ll-fitting with our environments. Sometimes pain is triggered by errors or blips in the complex entanglings of our bodies. But most of the time there’s a valid trigger.
One part of suffering is the manifestation of pain. As we need pain to stay alive, suffering is the price of life.
If you want to go Buddhist, it’s also the gap between desire and reality. The world we live in appears to have regular physical laws that constrain our activities. We also exist in something time-like, meaning change is inherent in all things. We also live in a social world with other people and other beings, all other fizzing processes attempting to locally beat the laws of entropy in a universe that collectively bets against them.
A better question is maybe: what would happen if we didn’t suffer?
If we had no pain, we’d quickly die, mangling our bodies and minds, placing ourselves in situations that act to destroy us.
Even if we had all we want, we would have to have all we want at all times, where our wants change over time (satiety and all). Despite the extreme physical difficulty of this, those that have come closest to this state are often miserable. The emperors, the kings and queens, the film and music stars, the Indian and British princelings, the trust fund children. Suffering provides a narrative gradient to answer questions 1 and 5, without it our existence seems superfluous. We cannot reach our goals completely.
ChatGPT:
7. What is consciousness and how does it work?
Me:
The standard answer to this: that we don’t know, that it’s all a big mystery, is increasingly bollocks.
The last 20-30 years of neuroscience research has provided a veritable bounty of information derived from experiment and careful observation.
For a good introduction, read Anil Seth or Daniel Dennett or Robert Sapolsky.
A key insight is that there are multiple degrees of consciousness. The unconsciousness of anaesthesia is different from the unconsciousness of deep sleep which is different from the unconsciousness of rapid-eye-movement (REM or dream) sleep which is different from different forms of comas.
What unites the most “conscious” states of consciousness is an access to our (processed) sensory impressions. Magic and comedy work off the difference between conscious and subconscious awareness. Consciousness also shows a particular patterns of synchronised brain activity across many different brain regions. But not synchronous in the form of predictable regular patterns (that’s deep sleep), more synchronous in the form of free jazz – there’s a tune and a melody but different players get to lead and follow.
It looks like it works by allowing us to flexibly generate new action plans of behaviour based on objects (physical and mental) within a current work space. We need to consciously repeat a sequence of actions with effort to learn a skill that, over the medium term, embeds below conscious awareness. Without it we would flit from automated response to automated response. That is likely fine if the environment is relatively stable. But in an unpredictable and varying environment (which is almost definite in even a small social group), we need to quickly change and adapt learned procedures by composing different parts together in a modular manner. This is what consciousness does.
ChatGPT:
8. What is the secret to happiness and fulfilment?
Me:
First, there’s no singular “secret”.
Second, it’s not really that secret – longitudinal studies, religious texts, and modern psychology shows us pretty clearly what the answers are:
- social engagement – family and friends, partners;
- working on collective projects with kind & friendly small teams, where the scope of the projects are easy to define, reasonable to complete on a semi-regular basis with fair struggle, and easy to integrate into a personal narrative of life’s journey;
- health of body and mind – only really seen when removed – to obtain it the old adages of healthy eating and regular exercise need to be followed;
- gratitude – there will always be wants, which are always relative – comparing with those worse off and being grateful for your advantages results in happiness, the inverse makes you miserable; and
- having enough – lacking a base level of shelter, food, heat and other biological necessities will naturally lead to a dissatisfaction that drives towards life.
ChatGPT:
9. Is there other intelligent life in the universe?
Me:
We don’t know. It’s an empirical question. We might know one day.
My hunch is that life exists elsewhere in the universe. The recent detection of thousands of exoplanets makes it seem like solar systems are common. The mind-boggling scale of the universe and laws of probability suggest that self-replicating systems of some kind do exist, and are possibly semi-common.
The question of whether “intelligent” life exists engenders two further questions. One of intelligence itself; the other of time and space.
History indicates that when we use the term “intelligence” we mean exhibits behaviour similar to the dominant cultural norms. We think ChatGPT is clever and serve octopus as a starter. So the question we are really asking is – does life sufficiently similar to the physical and cultural context of the question asker exist? The answer to this is: it’s quite a bit less likely. This requires a match between two or more multi-billion year sequences of actions. We still can’t talk to magpies.
This brings us to the second question, and necessary caveat: does that intelligent life exist within a spatial-temporal overlap that allows the question asker to communicate with it? This brings us into questions of technology that can bridge space and time – without it, it is very unlikely that in the huge void of time and space sufficient overlap would exist. At least before both sets of “intelligent” life annihilate themselves.
ChatGPT:
Err – “the discovery of microbial life on Mars”?
I guess this comes from this – https://www.amnh.org/learn-teach/curriculum-collections/cosmic-horizons-book/fossil-microbes-mars. This is from 2000 and I believe has mostly been discredited as evidence. Wikipedia says:
Much controversy arose over this claim, and ultimately all of the evidence McKay’s team cited as evidence of life was found to be explainable by non-biological processes. Although the scientific community has largely rejected the claim ALH 84001 contains evidence of ancient Martian life, the controversy associated with it is now seen as a historically significant moment in the development of exobiology.
https://en.wikipedia.org/wiki/Life_on_Mars
10. What is the nature of reality?
Me:
Again, the last 150 years have given us ever detailed answers to this question. The real question is often: is there a description of the nature of reality that nicely fits the narrative models of most human beings? (Especially those with limited maths skills.) The answer to the latter question is increasingly “no”.
The big leaps forward in our understanding of reality in the twentieth century were in the realms of the very large – relativity – and the very small – quantum physics. We’re still struggling to glue these two developments together in any meaningful way.
Carlo Rovelli writes beautiful short books on both areas. The nature of reality is a poem.
Both developments point away from any absolute nature of reality. In the sum, we are left with interacting local frames of reference on multiple scales. Time may be the interactions of shuffling pieces of a discrete machine that springs in and out of existence.
The term “nature” is interesting. What is the “nature” of reality? Not: what is reality? Used in this way, we expect some kind of distilling of what we know of the world. We don’t want Borges’ scraps of maps lying over the world. The term is also used when we talk about “the laws of nature”. There we use “nature” as a synonym for reality. So why the tautology in our question?
We also use “nature” to refer to the living world: Mother Nature; the natural world. “Reality” is dryer and non-living. Does that mean in our question we hope that within reality there is some guiding force of life? Do our questions give away what we hope to see?
ChatGPT:
Observations
The answers to these question in combination seem to suggest that ChatGPT is being driven, or has learnt, to follow a rough template of a response. It has obviously been tweaked to avoid anything that could be politically charged. This however leads to answers that feel a little too washed out, with few tangible and useful elements. The process of averaging removes specific information.
We also learn something from the questions themselves. The averaging over all the Internet.
We see that many questions actually have a fair few agreed upon answers. But it turns out we’re not really interested in those answers.
I often wonder whether we feign interest in the “big questions”. What we actually want are easy answers to the practical questions of our daily lives: How can I be happy? How can I find a partner? Where’s my next meal coming from? How can I stop my partner doing X? How can I earn more money? How can I easily feel good with no strings attached?
We also want answers specific to our context and constraints. We don’t care for a general answer, we want an answer that “gets” us and how we could and would act. But also we love fiction and gossip, so we also want specific answers for the context and constraints of well-defined others. They only need to be “true” in the sense of not contradicting our known (or yet to be known) experiences of life, not “true” in the sense of material presence in the world.
Robotic Ape 