ChatGPT vs. TRUTH
Discover more from You are the sky now
What’s the difference between humans and A.I? The relationship each has with truth (factual) and TRUTH (transcendent).
In a digital world that blends actuality and misinformation, A.I is truth (factually) agnostic. Referring to TRUTH defined by morality or spiritual realities, A.I remains far removed from lived, human ideologies.
Looking at interpretations of truth can reflect both parallels and vast chasms of difference between the ways humans and A.I (specifically ChatGPT) present their ‘thinking’ in words. Our very humanity – emotive and experiential – is the one thing A.I can’t replicate.
Truth and TRUTH
The search for TRUTH sets our values on our moral compass. It seeks something ‘bigger’ than us through awe and actualisation. I’ve written before; Our relationship with truth is what makes us human,
“Truth is important. It allows us to succeed. Both the facts (the actuality at hand) and our own personal truths (the values we live by) help us make sense of the world and move forward.”
Representative alignment issues
Facts and actuality at hand aren’t subjective, but generative language models reflect back content that isn’t always right.
We know that A.I reflects a scrapbook of existing online content, with the ‘glue’ being predictive language models to fill any gaps. There are copyright, accreditation and plagiarism issues here – Github is currently suing ChatGPT for using some of its code.
The complex, long-term issue of information bias also exists. Copy isn’t checked for sensitivity or nuance (although there are ongoing projects working on this). It’s modelled on the ‘best content’ it can source.
Logic would have it that these sources are fed by populist, advertorial and profit-orientated, human-made algorithms. Theoretically, factually correct can be surpassed by factually more prevalent. What happens in the black box, between input and output isn’t clear.
A critical eye
What is clear, that Chatbots like ChatGPT will increasingly feature as an automated part of our lives. There are tons of features on how to edit A.I-produced work. If we’re setting test tasks in the workspace or academia that can be done by ChatGPT, that’s a prompt to think differently. The most meta (and homogeneous) of all brands is one with a tone of voice written by A.I alone.
If you’re looking for the linguistic equivalent of what can be produced by a ChatGPT, you’ve failed already. That isn’t a true measure of academic or humanistic value although granted, it's an impressive imitation.
We need to reevaluate how we use words, to write stories and encourage deeper, critical thinking alongside fact recall. The point of difference here is that humans respond to a different brief. They write the brief. Our best writing plays with layers of meaning, nuance and TRUTH at a higher level.
The point of difference here is that humans answer to a different brief. They write the brief.
SPOILER: ChatGPT doesn’t have feelings
I asked ChatGPT for its own spin on truth and A.I,
“Artificial intelligence systems do not have feelings or emotions, so they do not care about truth in the same way that humans do. However, it is important for artificial intelligence systems to be designed and trained in a way that allows them to accurately and reliably process and utilise information. This requires that the data they are trained on and the algorithms they use are based on factual information and that they are able to accurately evaluate and interpret the information they are given. In this sense, truth is important for the proper functioning of artificial intelligence systems.”
What does this tell us? Truth here is important for ‘proper functioning’ of a generative language model but not in the moralistic sense. Your query results aren’t born of nefarious intent. ChatGPT is delivering what it’s been designed to do.
And yet for humans, truth is linked to morality. ChatGPT doesn’t pretend to be truthful in the need to convince anyone. There’s no weight of external judgement or a need for validation. It may ‘refuse’ to answer controversial topics, but exists outside of any human moral code itself.
If we’re looking for truth or collective accountability from ChatGPT, we’re holding it up to morals that it doesn’t have. It simply shuffles our digital universe and deals it back to us. There’s no ethical persuasion card. So, when you gaze into that abyss – or say, ask it how to calculate your pre-2020 carbon footprint in the style of David Attenborough – the best reflection of a factual truth (or a possibly distorted version) comes back to you.
ChatGPT is busy
What’s so convincing (and arguably unnerving) is that ChatGBT uses language that’s smart and conversational. I checked in on it this morning during a ‘high demand’ maintenance phase. It gave me an Upstart Crow-esque brush off – zoom in if you’re a fan,
Perhaps that was my cue to step away from the machine. Because the innate thing that exists in the non-virtual world – diversity of input – nuanced, experiential input that gives us an enhanced output – is uniquely human.
The creatives? The doers? Life is their reference.
ChatGPT gives no weight to TRUTH in the existential or ethical sense. An automated definition yes, but not the higher understanding or its moral implications. In its own words, it doesn’t have feelings or emotions. It can’t comprehend why you may feel hopeful, threatened or conflicted about it.
The truth that A.I aligns with is facts. Although we need to work on getting those right first. On the plus side, ChatGPT is open about some of its weaknesses. It explains it can make mistakes and is only as good as its inputs – a self awareness shown too by wise humans.
Ultimately ChatGPT can’t get outside of itself. It can only give us a theory for how we might try to ourselves. Living is TRUTH unique to humans.
Experiential understanding? That’s for us to do.
More hot takes on ChatGPT
A.I and ethics (plus the mirror metaphor), Forbes
A.I, copywriting, coaching and running, Bandit running
Chat with Gary Marcus, Professor of Psychology and Neural Science at N.Y.U about A.I, The Ezra Klein Show
Is ChatGPT the future of humanity? Reasons to be Cheerful