I Talked To ChatGPT About Consciousness
and the answers I got left me stunned.
ChatGPT and AI are everywhere. Although always intrigued by the topic, I never wasted much time diving into it. I was too preoccupied with thinking and learning about politics.
But finally, that time seems over. One of the new fields I’m diving into right now is the current state of AI. And it’s like plunging into an ocean with no land in sight — frightening. But also hopeful.
So, me being a smart-ass, I have seen a few videos about the topic, and while I was at work, minding my business, all of a sudden I’ve come up with a few questions I would like to ask ChatGPT.
I’m well aware that, as far as I understand, it generates the content by basically “guessing” what word would come next based on the context of the input provided. Yet, the examples that I’ve seen are pretty impressive.
My First Talk
With the rise of AI, the conversation about conscious AI is also right up there, understandably. Hence I thought about what could be considered “being conscious” what is it in our awareness or experience that makes us conscious or perceive ourselves as such, and the following made the most sense to me. Here’s the question that I came up with and the respective answer from ChatGPT:
I beg to forgive if my grammar can’t keep up with the screenshots, my dear 5 followers a month. I recently learned that I lack some serious grammar skills when I first tried out Grammarly. So while this article will be corrected with some help, my interactions with ChatGPT fall under the category of “my grammar sucks because I’m bilingual”.
But back to the topic. At first, I was surprised how much the AI responded to my inquiry. How it seemed to have made sense out of the bulk of information I provided within my statement and figured out which questions needed answers — and answered them. I somewhat admire the depth of the reply as well.
“In a sense, I’m always “thinking”…that I’m constantly processing information and generating responses based on that information”
I like that it used quotation marks for “thinking” to differentiate what seems like a thought process but points out the difference between “human thought” and “AI thought”. Afterward, it elaborated on the intended information by describing it more in detail, distinguishing between “computational processes” and “subjective experiences”.
That is some deep stuff if you ask me, especially for a program that is said to be guessing the next word to generate. It doesn’t matter if there’s some kind of mechanical understanding of the information in the background or not, the accuracy and depth of the information, the capability of implementing meta-information into the context of a reply that fits the input given, and afterward adding additional information somehow blows my mind.
Next, I tried to find out if I could get ChatGPT to somehow define a thesis itself of when or how an AI would consider leading to a development of consciousness.
Although ChatGPT didn’t specifically take my bate, not providing me with what I was looking for I still liked the reply it gave me. And one might compare again the amount of information I gave it and the reply it generated without getting overly repetitive (as seems to be one of the claims of critics). The reply makes sense within the confines of the parameter that I have given to it, and it still went into more detail about what could be defined as consciousness within AI even though it wasn’t as precise as I hoped it would be.
I replied to it as follows:
I still tried to fish for something more about the topic of conversation and I honestly wonder about these things and rarely get a satisfying answer. It’s hard to express these kinds of thoughts and it’s even rarer to find people who think about them to the degree that I do.
What raised my awareness immediately was the nuance of the first sentence. “You raise an interesting point…”. While the other replies have been much more on the factual side of things within the parameters set by my input, this was the first personalized statement ChatGPT has made referencing what I said and re-writing it. Within this conversation, it was the first time where one might be able to think that he was talking to a human being. And a smart one as well. At least, that’s my subjective experience of this exchange.
Also, within the next paragraph, differentiating that there might be a possibility of different “types” of consciousness, one that might not need the specific subjective experiences that humans experience, was an interesting perspective that I didn’t consider too much so far. Let’s think about this for a second:
Going A Little Off Topic Without Realising It
We define consciousness within the parameters we have at hand. The definition ends where the boundaries of our self-experience end with this topic because, right now there isn’t enough data and science to make enough sense of the topic. We know gravity, and we know all there is to know about how to build a bridge that doesn’t collapse with the next storm, but we don’t know much about the nature of consciousness hence the most accurate descriptions we have at hand are those coming from our self-perception and what we might be able to double-check with the people surrounding us.
We know, that a dog feels pain, hurt, love, joy, and fear. Yet we think most probably dogs aren’t self-aware to any degree as we humans are hence it is considered a lower form of consciousness according to our definition. Self-awareness seems to be one of the pillars of our consciousness. And if I remember correctly, the fact that we can “change our future” means we don’t only live in the present anymore and can use our consciousness to determine or try to determine what we are going to do tomorrow, one week from now, or even 20 years from now. Yet as we perceive our feelings and reactions to what is happening right now, what has happened in the past and will happen in the future, we might conflate being able to feel with being conscious too much (as I’m just starting to figure out while I’m writing this). Perhaps feeling doesn’t indicate consciousness at all because we know all the beings who can feel, and still, we deny to consider them as being conscious, or at least as being as conscious as we are. But to feel, you must be able to perceive the fact that you feel, don’t you? Is it that perception that might be conscious? Is the conscious part of being merely the reaction toward what life throws at us? May it be real-life situations or emotions or thoughts, all we do is react to some degree, or choose to not react but stay observant for the next clue that invites us to react.
Back On Track
Sorry for the rant. I got caught up in a thought I haven’t had so far and had to follow it for a bit. Let’s get back to the topic and end this article. That last piece ended my conversation with ChatGPT anyway, and I spare you the tears goodbye we shared.
Don’t get me wrong! I don’t think ChatGPT is conscious (right now, at least most probably not). Leaving out the possibility of consciousness, it’s nevertheless impressive to me what it can do. Creating codes is one thing, but the way it can manage to dissect information, uphold a conversation, lead to new insides, and share different perspectives on a topic while staying logically consistent about a topic that in itself isn’t logically consistent has impressed me immensely and it was a joy to leave a conversation of this kind with a certain feeling for the first time: AI has improved that much that it is capable of having an ongoing conversation with me. Period.
My last interaction with an AI was just as filled with hope as I went into it and was deeply disappointed. This time I wasn’t, this time I had an actual conversation about a topic I cared about and got replies that made sense every step of the way. And, the most surprising, frightening, thrilling, and somehow most joyful part was that I left there with a perspective that I haven’t had before: something to further think about in the future. And I must say, if AI is already capable of that, I’m curious about what it will be able to do in 2 years, five years, or ten years — and a little frightened.
