The Prompts That Bind Us, Part 1

In Part 1, I look at the idea of interconnectivity, highlighting how communication plays a uniquely pivotal role in the growing connection between humans and artificial intelligence (AI). With close to 135,000 years’ worth of experience in communicating with one another, human beings are now exploring the rising potential that AI has to create quantum shifts across all facets of society.

I’ve always been fascinated by science and was very lucky to have had an uncle who was a computer scientist who never outgrew his inquisitive nature. One of the things he introduced me to as a kid was the concept of quantum mechanics. What resonated most with me was its notion of entanglement; that everything in nature was somehow connected. Granted, that’s a highly reductive way to describe quantum mechanics, but it suited my ten-year-old imagination just fine. But this article is not about physics; I wanted to lean on a particular portion of that theory as it relates to interconnectedness. Now, theoretical portions of quantum mechanics do have a metaphysical cousin (if you will) that is usually found in the realms of philosophy or the paranormal. But regardless of the school of thought, I thoroughly enjoy the fact that they each share one basic principle within their respective dogmatic opinions: Everything is connected.

Given that notion of connections, I wanted to write about one type in particular that I see as the latest and possibly the most controversial connection in human history: The human/Artificial Intelligence (AI) connection. Go ahead; allow the images of HAL-9000 from 2001: A Space Odyssey to bounce around in your head for a moment. They’re bound to resonate with you by the end of this piece.

As social creatures, we are very attuned to the importance of communication. From hand gestures to facial expressions, protolanguages to writing, we have sought to expand our ability to better communicate. Throughout our long evolutionary existence, the quality of our efforts has risen to represent more complex and nuanced actions, activities, thought processes and consequences – whether for one-on-one exchanges or global connections involving hundreds of languages and contextual meanings. When computational machines came into the picture, we created machine languages based on binary code and symbols, slowly moving our way up to languages that used basic grammar rules, paving the way for the research that eventually would lead to AI.

For this article, I’ll be sticking to generative AI (GenAI) models such as ChatGPT, Claude, Gemini, etc. These use a neural network to process statistical analyses and pattern recognitions against huge datasets in order to generate creative content using text and images that would appeal to us organic types. The manner in which we usually communicate our requests to an AI is through prompts. These are descriptive requests you type up using a narrative format in your own natural language. The AI’s cognition comes from using massive amounts of natural language data so it can understand what you type and answer back in a very conversational fashion, regardless of the complexity of your prompt.

Now here is where things get a bit… fuzzy.

In today’s world (with the implied today having an incredibly limited shelf life), AI models are growing exponentially in terms of capabilities and functionality.¹ Yet despite the sophistication of their output and overall abilities for comprehension, none are considered to be truly self-aware or sentient. It all depends on their continued training. And that includes what they learn from their interactions with us.

Sure, they do learn from those massive datasets. But it’s their interactions with people that help refine their learning. When addressing this type of training relationship, the tech industry explains how this interaction works as follows: “it plays a role in refining and directing that learning (from datasets) to better align with human preferences and values.”² Granted, this is a relationship that is constantly evolving, especially when you consider the number of people that these AI models interact with globally on a daily basis.³ And in regards to human-AI learning, the tech industry points out that these interactions enhance “the (AI) models’ ability to understand, generate, and interact in a more effective and helpful way.”⁴

Let’s take a quick look at some of the words I highlighted from those two quotes: values, understand, generate, interact, effective, and helpful. When you think about it, these are the same descriptors we would like to apply (or should) to the conversations we have with one another. But consider the following:

Do we attribute these terms to our general conversations? Do we always aim to add value; to understand one another, generate interest, or interact in an effective and helpful way?
Do we conduct our conversations with an aim towards learning something new, or to share a piece of information or insight that might help teach others regardless of importance or relevance?
Do we communicate primarily in our daily conversations to exchange, explore, or express?

That last question is perhaps a bit ambiguous. Many of our daily communications have moments where we exchange information with more than one person. On other occasions, we may explore ideas, concepts, or potential solutions that would aim to teach or be used for learning, be it in a one-on-one exchange or with a group. And I think it’s safe to say that every conversation we have aims to express something, whether it’s an opinion, an idea, an emotion, or even the subliminal underpinnings for a motive or a point of contention.

So regardless of how or what we communicate, each exchange between us organic bipeds is usually rich in meaning, content, and purpose. They provide a tapestry of intellectual interconnectedness that helps to link us together as a species.

So, What About AI?

Well, artificial intelligence is not sentient (yet), and neither Artificial General Intelligence (AGI) nor Artificial Superintelligence (ASI) exist in today’s current technological landscape. These types of AI are conceptual levels currently being studied that are expected to equal or surpass human intelligence.⁵ They will be imbued with the ability to understand, learn, and apply their new knowledge across all aspects of problem-solving, creativity, and even general cognition. So if we look at the spectrum of where AI was just a few years ago, where it sits today and where it may be in the near future, we need to ask ourselves: How are we communicating with AI?

Here’s where the quandary of interconnectedness comes into play as it relates to AI. As human beings, we learn to communicate from the moment we’re born; abilities such as recognizing language, tastes, and emotions are learned while in utero. It’s a simple yet straightforward process for learning that establishes foundational preferences that will guide us in those formative years as we continue to grow and our mental capabilities expand and cognitive skills increase. If we think of AI’s coming of age⁶ as a birth, then it too began learning while in utero. By being fed massive datasets by highly skilled engineers, the fledgling AI was being provided the means to understand how to navigate the use of its computational tools while residing inside a complex data center or server farm.⁷ This would eventually lead the AI to create content that we could then visualize and use for consumption, research and learning. And when the AI’s neural pathways were ready, the AI was born into our world. Its infancy progressed at astounding speeds, learning more with each interaction it had with its users located all around the world. But what has AI been learning from us? If you recall those words I pointed out earlier, do any of them represent the manner in which we approach our communications with AI today?

being frank