When I was younger, I was a voracious reader. This filled my vocabulary with unusual words that many people around me had never heard. These words gave me a significantly different perspective from everyone else. When we look at studies about brain growth from learning other languages, we see that it promotes different neural connections, perhaps even enhanced capabilities for lateral thinking or intuitive processing. In this way, new words can fundamentally reshape how we perceive our world. I didn't have a second language growing up, so I spent my time pondering words.
One of the beautiful aspects of exploring new words is that they arrive without any pre-existing context. While we must be conscious of the surrounding material that our words will eventually attach to, this blank slate gives us a unique opportunity. We can expand and explain this new concept to another person, potentially opening their eyes to seeing an unspoken idea in an entirely new light.
Let me provide an example with the word "compersion." Compersion is often described as the opposite of jealousy. It's a polyamorous relational term that describes the ability to feel joy when someone you love finds love in their life, even if that love isn't directed toward you. It can also be understood more simply as a form of empathy – feeling happy when someone else is happy. Your love grows alongside theirs, even if their happiness comes from someone else or something else entirely, like a hobby or passion.
When I was dating polyamorously, the word compersion opened up an entirely new dimension in my emotional relating that I simply hadn't considered before – largely because I'd never encountered the word. The word acted as a banner under which I could rally my understanding and transform how I perceived my emotions. I suddenly had language to explain my feelings: a word that enabled new behaviors and helped me reframe my emotions in a way that made sense, rather than leaving them in that cloudy, murky, vague space where many emotional subjects reside. I evolved from being a passive consumer of my emotions, sometimes even their victim, to someone actively creating and understanding my emotional reality. Having this new word was genuinely empowering.
One common phrase I found people would repeat, though I didn't necessarily agree with it, was "relationships require compromise." Friends and well-meaning people told me this repeatedly, but it always struck me as strange. At first, I dismissed my doubts, thinking I was simply being stubborn and laughing at myself for questioning the accepted wisdom. But as I considered it as a fundamental truth about relationships, I found the foundation surprisingly shaky. I began to examine the word "compromised" itself – how we use it in phrases like "I compromised my values" or, in military parlance, "our position is compromised." I came to see compromise as essentially a compromised word – a word that detracts from our enjoyment of shared relationships. To gain clarity around this concept, I realized it was an excellent opportunity to create something new.
My goal was to develop a word that would serve as the opposite of compromise. In this new concept, nothing would be lost – instead, both people would grow in harmony with each other, rather than sacrificing parts of themselves.
I realized this word needed to incorporate some form of agreement at its core. Often in relationships, when we say "compromise," what we really mean is agreement or negotiation. Since relationships are fundamentally built on agreements, this seemed like a much stronger foundation. I also recognized that change is inevitable in relationships – our bodies physically transform as our cells regenerate, making us new people in a very literal sense. Our relationships need to evolve and update as we change, fostering growth rather than stagnation.
While agreement formed the base, I saw that we also needed to incorporate empathy. I came to understand that compromise often stemmed from empathy – we weren't really losing anything in our relationships; instead, we were gaining deeper understanding and connection with another person. The apparent "compromises" were actually opportunities for expanding our empathy and emotional capacity.
Additionally, I wanted to acknowledge the profound impact of the other person in the relationship. Their love, attention, and presence all contributed to this agreement to change, directly influencing our personal growth and evolution. They became active participants in our transformation rather than obstacles to overcome.
Having recognized the need for a new word to encapsulate this concept, and having assembled all the necessary components, the next challenge was combining them effectively. This is where AI proved invaluable, offering non-intuitive ways to connect these different concepts.
My discussion started over voice chat with a search for a word that's the opposite of compromise. ChatGPT provided a few examples.
Ed: “I need a word that's like, it's the opposite of compromise, but it's not like, it's like adapting to someone joyfully, rather it's compromise, you lose something. This is like adapting to something joyfully. What would that word be?”
ChatGPT: It sounds like you're looking for a word that captures the idea of willingly and joyfully embracing a change or adapting to something with positivity. The word "embrace" might fit well here, as it conveys the idea of accepting something willingly and with enthusiasm.
Other suggestions it provided were adapt, concord, concur, enthuse and harmonize. However, this wasn't fitting what I wanted. So I asked:
Ed: “Can you take all these words that you've suggested and I've suggested and brainstorm, perhaps 10 words that might be combination of words or new words that people would hear and immediately understand.”
ChatGPT: Absolutely! Here's a brainstorm of 10 words that might convey the positive and joyful acceptance you're looking for, using combinations or new formations:
- Joycept
- Blissapt
- Embracement
- Cheercept
- Joycord
- Blissgree
- Enthcord
- Joyousent
- Blisscord
- Cheercept
These words aim to blend concepts of joy, agreement, and acceptance to convey a positive...
Ed: "Now think of words that are formed more like compersion." Accessing a latent space here.
ChatGPT: Got it—words like "compersion" often combine a Latin or linguistic root with a suffix, typically designed to evoke a specific feeling or concept. Here's a brainstorm of new words following a similar structure:
- Concersion – Derived from "con" (together) and "version" (turning), suggesting a joyful turning together.
- Emotersion – A mix of "emotion" and a suffix implying sharing or flow, reflecting shared positive feelings.
- Adaptension – From "adapt" and "ension," meaning to joyfully adapt to or harmonize with change.
- Accersion – Rooted in "accept" and "ersion," conveying joyful acceptance and growth.
- Harmoversion – A blend of "harmony" and "conversion," symbolizing a shift toward joyful unity.
- Resonersion – Inspired by "resonate" and "ersion," emphasizing harmony and shared joy.
- Affersion – Based on "affection" and "ersion," reflecting a warm, joyful connection.
- Universion – From "universal" and "ersion," expressing shared joy in unity.
- Sympersion – Combining "sympathy" and "ersion," signaling shared positive feelings.
- Amicersion – Derived from "amicus" (friend) and "ersion," highlighting joy in friendship and unity.
From here I decided I liked Sympersion, and asked it to expand with the etymology, after which I workshopped the definition. It's worth noting that many of the AI's initial suggestions weren't particularly strong, but that wasn't the point. The value lay in its ability to try different combinations and lead us down interesting pathways, evoking new concepts through our interaction with AI.
Sympersion, a definition
Sympersion (noun):
A state of harmonious co-regulation and mutual growth in a relationship, arising from empathetic growth and shared agreement to change. In Sympersion, each person acknowledges the profound impact of the other's love, attention, and presence on their personal growth and adaptation. Rather than perceiving change as compromise and therefore a loss, it is understood as an expansion—a growth in who each person is and what they contribute to the relationship. Sympersion represents the dynamic flow of emotional attunement and physiological alignment, where partners navigate challenges and changes together, fostering mutual resilience, connection, and transformation.
"Sym"
Derived from the Greek syn-, meaning together, with, or united. This emphasizes cooperation and shared experience, which aligns with mutual regulation in relationships.
"Sym" also evokes the sympathetic nervous system, creating a more direct connection to the physiological aspects of co-regulation.
"Persion"
Likely derived from persevere (to endure or adapt through challenges) and persuadere (to move toward agreement or alignment).
Persion could also hint at perceiving (as in emotional and nervous system attunement) or dispersion, suggesting a flow or spreading out of balance and harmony between partners.
Prompting
Since language models could be described simply as pattern matching devices, it is extremely good at this type of task. Many people think that AI is not good at creating new things. But in fact, the opposite is true. Language model models which use predictive word patterns are able to see across a large number of concepts and bring them together. Their ability to synthesize and blend different word meanings is remarkably powerful.
However, there are important caveats to consider. Since these models operate by predicting the next word using statistical probability, their initial answers often tend toward the bland or conventional – simply using what's most likely to come next. The key to getting more interesting results lies in prompting the model to make unconventional connections.
I think this is where a lot of dissatisfaction amongst people using language models has come from. They go in, they use things, and they find the answers are sort of like something you could just get by Googling or going on Reddit. A great example of this recently was someone was asking "What are the top five books that you would recommend to change someone's life?" And the answer was like "Atomic Habits" and so on. Simple personal growth type books that were best sellers. And it's picking those because those would be the most popular choice. So when you are using language models, consider that the first result you will get if you don't prompt cleanly in an interesting way, you'll get the top 40 types of results from this language model radio.
This limitation explains much of the dissatisfaction people experience when using language models. They often receive answers that feel like something you could find through a simple Google search or Reddit post. A recent example illustrated this perfectly: someone asked for the top five life-changing book recommendations, and the model responded with predictable bestsellers. The model selected these because they represented the most popular choices in its training data.
When using language models, it's crucial to remember that without creative prompting, you'll often get "Top 40" results – the most common and popular responses from the model's training data. It's like tuning into a commercial radio station that only plays the current hits, rather than exploring the full spectrum of musical possibilities.
One fascinating way to think about language models is to perceive them as vast landscapes or spaces to explore – latent space. Think of latent space as a 3 dimensional space where similar concepts cluster together. When a language model processes text, it converts words and phrases to "vectors" (points) in this latent space, allowing it to capture nuanced relationships between concepts that aren't obvious in the raw test.
For example, in this space, the word "king" might be positioned in relation to "queen" in a way that mirrors how "man" relates to "woman" - capturing not just similarity but analogical relationships. The fascinating thing about latent spaces is that they emerge from training, rather than being explicitly programmed. The model discovers these meaningful representations on its own.
This rotatable 3D visualization demonstrates how language models navigate semantic relationships in latent space. Mouse wheel or pinch will zoom.
I envision language models existing in a latent space that contains vast valleys, islands, deserts and mountains of different landscapes formed from their training data. When you prompt these models, you're essentially accessing different regions within this conceptual space. Latent space is a slightly controversial subject around language models. However, in the image diffusion space which I know quite well working with models like Stable Diffusion due to my work, I find that there is definitely a latent space for trained materials and it can easily be accessed depending on your prompt. To me, it doesn't seem like as much of a leap to perceive this internal world of language models as a latent space.
To navigate the expanses of latent space and experiment with words, I've found it particularly illuminating to leverage language models' visualization capabilities, like their ability to generate Mermaid graphs. A Mermaid graph is a simple way to make diagrams or visualizations using text descriptions, whether they be flow charts or user journeys. There are all kinds of different capabilities you can access using language models that can code, using Anthropic's Claude language model you can create code artifacts which enable you to visualize concepts relatively easy.
Example of a simple flow chart using mermaid graphs.
In one experiment, I prompted Claude to envision a rather unconventional scenario: a dialogue between a Buddhist monk and Emily Dickinson exploring spiritual transformation, through the language of jazz. I then asked it to map the emerging lexicon in graph form. This might sound absurdly experimental – and it absolutely was – but such creative extremes often yield the most intriguing linguistic discoveries.
How this relates to interpersonal communication is in the way our words translate across the membrane of our consciousness. We carry latent meaning associated with our words, similar to how a language model has latent training data surrounding the information encoded within it. We hold emotional connections, attachments, experiences, and memories, all associated with specific words that give them richer context in our minds when we use them. However, others may not share these associations. They might have received different cultural conditioning regarding particular words, or they may have had different experiences or memories associated with them.
In this example, I've used the LaTeX scholarly format as an output from the language model to make an essay on some of the words explored through the process. Because everything looks far more official when you put it in LaTeX.
The interesting thing about this is in finding out how our words translate across the membrane to the other person. Sometimes it requires deep inquiry, asking questions about how they perceive specific words to fully understand their nuanced interpretation of our intended meaning. When we can create shared understanding, these words enable us to explore new spaces in our mutual inner worlds.
Sympersion recognizes the unity that emerges from our civilization. There is no compromising of oneself within the unity of the whole. As members of one organism—the human race—we act in concert with one another. Consider a symphony: the bassoon player does not say, "The violin soloist is taking up all the attention, therefore my part is not as important." Everyone has their essential role to play. We do not view change as something that negatively impacts us; instead, we embrace it as a source of empowerment.
This embrace empowers us to see the bigger picture in our lives rather than become entangled in self-righteous defense mechanisms. We may find Sympersion in how we relate to artificial intelligence, or in how we relate to nature as climate change increasingly affects our society. Sympersion could serve as a valuable tool in navigating spaces of shared agreement and expansive empathy, rather than focusing on the limitations implied by cup half empty of compromise.
As we venture into unprecedented territories that AI is revealing to us—these latent worlds of understanding that comprise the fabric of our reality—we will need new words and new ways to perceive the world. Let us create them with intention.