The Deterioration of Communication: Navigating Contagious Meaninglessness

Intimate conversation has ancient roots, but it has also become a rare treasure in today’s society. How do us technologists help fix that?

Quick Note: If you’d like to listen to this post instead of read it, check it out on Substack, where I post weekly with voiceovers. Find it here.

Yeah, we’re pretty much talking about language models again. Don’t worry, that’s not all we’re going to talk about on this blog. However, language models are powerful in many respects, and are capable of a variety of novel complex tasks. That means they’re worth talking about. We now have automated language. That means the effects of this technology will be felt everywhere that language is used. This especially means writing.

The marketing teams at big companies have been hard at work selling us potential benefits of the new technology. Think of the ways that automation will benefit society! Never again will humans be slaves to an employment machine! Think of the liberation that will be felt by a generation that doesn’t have to work! The world will change forever in the coming decades, and the future is bright!

However, us excited innovative types might take a bit of advice from the skeptics.

Haven’t you seen Terminator?! This never ends well…

As far as I can tell, the human-extinction-bent killer robot army is a problem for future me. There are currently more pressing concerns. Today, we’ll be talking about one of them. A degradation of a fundamental aspect of human nature. A degradation exacerbated by technological developments and incentive structures within our economy. Today, we’ll be talking about the collapse of communication.

Author’s note: this article contains some opinions. While I’d like to think that I’ve thought these things through, I’m also completely open to disagreement and criticism. If you feel compelled, please share your thoughts in a reply below.

The Digital Campfire

So back in the day, that is, approximately one to 1.5 million years ago, us hominids were just starting to play around with fire. Once we got the hang of that, having a place to keep the fire made sense. That way, the whole squad could gather around for warmth and protection. As the central meeting place for a given tribe, campfires became social hubs for hominids. Culture itself has its foundations in campfires.

Many of us are lucky enough to still find ourselves around a campfire from time to time. However, intimate conversation finds itself a rare treasure in modern society. A common sentiment has begun to form:

We’re simultaneouly more and less connected than ever.

There are a few reasons for this. One is that digital communication simply doesn’t convey as much context. You can have a conversation with someone online, and when they say things, they could be crying or fuming with anger, and very little of that emotion gets through. In person conversation allows for reading of cues that aren’t available in the lower-dimensional level of textual conversation.

Another reason that we feel less connected, is because online conversations are asynchronous. When I send a message to friends online, I often don’t expect any immediacy in response. They will reply when they have time, and then I will do the same. This means that oftentimes online conversations aren’t conversations where both parties are focused. Attention is simply not given to the conversation at hand. In person, conversations require mutual attention.

It’s worth noting that the digital landscape merely promotes these flaws, it does not require them. A good communicator can convey emotion through writing. Additionally, I have had many excellent conversations with online friends when we are both present and engaged in the conversation at the same time. Not to mention the benefit of being able to communicate over long distances, and engage in conversations over longer timeframes.

However, despite significant utility in the digital world, the flaws do exist. People are more lonely than ever. Our minds, evolved towards the campfire, don’t know how to deal with this new social environment. Is the influencer my friend? Do I have a relationship with them? Our brains are not good at telling the difference between reality and online depictions of it.

Paid Attention

By now, the term “attention economy” is likely one that you’re familiar with. Essentially, one of the most profitable business models in the world is that of putting eyeballs in front of ads. Google’s revenue is almost entirely composed of ad money, as well as YouTube, Facebook, Instagram, X (formerly Twitter), Snapchat, and TikTok.

The business model is sound: simply create a platform which many people use, make it engaging to maximize your users’ time spent online, and display ads based on their preferences. Then, you just need to find a place to store the money. It’s lucrative, but it works. However, there are dire consequences for mass scale engagement-driven social media.

Determining intentions

One key indicator of the intentions of a platform is the type of content that it promotes. For example, after spending time on X, you can expect to either feel angry or validated. If X shows you a post that you disagree with, then you might be compelled to engage. If they show you a post that makes you feel good about your own opinions, you might be likely to engage. Either way, it’s the extreme ends of the idea spectrum that are pushed, not those with nuance.

On Instagram, it’s a similar situation. People display a persona–an idealized version of themselves–rather than the realities of their lives. Buy why? This kind of content gets likes, as one side enjoys looking at it, and the other appreciates the viewership. This leads to engagement. Facebook does this too, along with TikTok and YouTube. All of them increase revenue by maximizing your engagement with their platform.

Attention theft

Engagement just means attention. They are just trying to get your attention. You might get something out of the experience, no doubt, these platforms have their benefits. YouTube has an amazing amount of high quality video content, and you can learn literally anything there. The other ones are cool too, I guess. However, think of the wasted energy consuming mindless content.

Of course, you’ve already thought about it. This isn’t news to you, and you’ve likely noticed this phenomenon of your attention being monetized for others gain. The alarm has been sounded for years.

Kill the Messenger

I’d like you to imagine the following situation:

A man named Bob must email a client, we’ll call her Jen. Bob does not like sending emails, as maintaining corporate professionalism is a task in and of itself. All Bob wants to say is this:

• the latest update is deployed, here’s the link
• Carl in IT says we’ll need to upgrade the server soon for performance
• can you send over the latest marketing assets?

However, this doesn’t have that sweet polish of a professional email. It contains all the content necessary, but none of the padding expected in the workplace. But Bob is a resourceful man. He turns to his favorite tool as of late, ChatGPT:

Hey ChatGPT, can you write the following as a professional email to “Jen”, from “Bob”:

[insert key points from before]

The language model responds

Subject: Latest Update Deployed & Upcoming Server Upgrade

Hi Jen,

I hope this message finds you well.

I’m pleased to inform you that the latest update has been successfully deployed. You can access it using the following link: [insert link here].

Additionally, Carl from IT has advised that we will need to upgrade the server soon to ensure optimal performance.

Could you please send over the latest marketing assets at your earliest convenience?

Thank you for your assistance.

Best regards,
Bob

Beautiful, now he can send his perfectly crafted professional email to Jen. In turn, Jen receives the email, but she’s a bit pressed for time. She knows there’s some corporate padding here, so she decides to get out her handy tool, ChatGPT:

Can you please summarize the key points of this email?

[insert email content here]

Certainly! Here are the key points of the email:

The latest update has been successfully deployed and is accessible via a provided link.

Carl from IT advised a server upgrade is needed soon for optimal performance.

A request for the latest marketing assets to be sent as soon as possible.

Equipped with her key points, Jen now must respond to her colleague… and you can take a guess what that looks like.

The future is now

This is the future. Email clients have begun to equip their users with the exact tools required for such a process. In Gmail, for example, you don’t even have to switch to another application for this functionality. Simply type what you want to say in the “Help me write” box, and through the power of generative AI, see an email manifest before your very eyes.

Additionally, Gemini sits atop every Google Workspace application, watching and waiting for you to ask it something. “Summarize this email”, or “List action items”, or “Suggest a reply”. Acting as the ultimate assistant, Gemini is available to prevent you from needing to lift a finger. But do I really need a language model to “Help me write”?

The situation we are creating for ourselves is one of personal messengers who act as superficial intermediaries between us and those we converse with. Intermediaries with the uncanny mannerisms of modern language models. Intermediaries, which widen the gap between us and our digital “connections”, and perpetuate the issues of online interaction.

Prophecy fulfilled

This future isn’t exclusive to professional emails, but could extend to all facets of online interaction. Here’s a headline for you from Futurism:

Bumble Founder Says Future of Dating Is Your AI Will Date Other People’s AIs and Hook You Up With the Best Matches

Oh, and yes, that’s a Black Mirror episode. She goes on to say:

“You could share your insecurities. There is a world where your dating concierge could go and date for you with other dating concierges.” — Whitney Wolfe Herd

There is an air of dystopia that surrounds such an envisioned world. If we’re disconnected now, adding another layer between real people can only exacerbate the problem. That’s not necessarily to say that online dating itself is ultimately detrimental. Dating is just one area of many that will see effects of online interaction-by-proxy. The future of social media might be one of AI “personas” who interact online and curate social media feeds of their users.

In pursuit of what?

I will say that I think this idea has potential for applications’ recommendations generally. Agentic interaction might be a good way to get some benefits of dialogical reasoning into applications. Determining compatibility is tricky, think about how many YouTube videos you scroll past before eventually finding one that sticks. Perhaps two personalized agents having a conversation to determine compatibility is a good way to increase the accuracy of a suggestion. Therefore, boosting engagement, and profitability.

However, in this future we must ask ourselves

“In pursuit of what, exactly?”

Is the point of social media to have an endless stream of compatible content to mindlessly consume? Is it beneficial to you, to future generations, to have such fascinating content readily available and perfectly curated? Of course, this is advantageous to the companies profiting off of your attention. As soon as individualized reasoning engines are profitable to run at scale for individual users, perhaps this type of recommendation will take hold.

However, we must recognize that such systems will likely fuel the dilapidation of intimate conversation. If the benefit of social media to the end user is social interaction, then disconnection from real interaction is ultimately detrimental.

Superficial love

From the same soil of loneliness of which online dating was born, a more corrosive trend has begun. Digitized love. Enabled by advancements in generative AI, relationships are becoming commoditized. Rather than embrace true human connection, paying customers can access fine-tuned, engaging, and idealistic glorified chatbots.

Replika, for example, is an AI company selling a companion. It started as a personal project by Eugenia Kuyda, where she used her text messages with an old friend to create a chatbot. Eventually, this idea turned into a service performed for others in grief, and eventually a full-fledged application powered by language models.

Now, you can create a “Replika”, which can act as a friend or other type of companion. The company markets their product with the following tagline:

The AI companion who cares. Always here to listen and talk. Always on your side.

Now listen. I’m not here to shame people for building some sense of parasocial bond with a lifelike chatbot. I get it, it feels somewhat like a person, and I’ve personally gained fairly useful advice through conversation with language models. They have useful information at their disposal. I have no doubt that there is some utility and potential benefit in having AI “friends”.

However, to say that this product “cares”, is nothing short of a complete fabrication. A language model has no capacity to care. It is simply not capable. The only thing that a language model can do, is approximate language. That makes it incredibly good at sounding like a person, but it also makes it completely incapable of feeling any human emotion.

To market an “AI Companion who cares”, is at best misleading, and at worse predatory. It can certainly seem like it cares, but there is no true meaning behind its words. Replika is one of many applications which market a companion who cares, or is capable of being an emotional outlet for a person. While they may be able to simulate real connection, all that’s happening is math.

The quasi-relationships built with these models have little-to-no meaning, and certainly are not the same as a human-to-human connection. Language models and generative art are not enough to build a person. You can mold a perfectly realistic bird out of clay and paint, but you cannot expect it to come to life.

As a sidenote, you can learn more about language models from my post about Transformers :)

Meaningless Content

On the topic of meaninglessness, the online writing industry itself is facing an infection. If you often read on the internet, you’ve likely stumbled upon a blog post that just felt a little bit strange. Perhaps it used the word delve one too many times, or just didn’t feel like a person would have written it in that way.

AI-generated blog posts have become one of the quickest and easiest ways to farm content. With a simple prompt and some examples, you can generate endless blog posts and newsletters in an almost entirely automated fashion. This system presented itself as an almost too obvious use-case for somewhat advanced language models quite early on.

However, as it turns out, language models aren’t actually very good at making blog posts that people want to read. As Matthew Ferrera over at Always Inspiring aptly put it:

In general, I cannot be bothered to read that which nobody has been bothered to write.

In other words, what’s the point of reading an article that no one actually wrote? This touches on some fairly deep levels of human psychology as well. We are social creatures, we care about other’s opinions, we attach weight to the thoughts of others.

A piece of art is a nexus of meaning which acts as an intersection between multiple dimensions of the creator’s unique experience. This nexus allows viewing from multiple perspectives, while still accessing the value imbued by the creator. This is why we actually find blog posts interesting, because they are in a way the culmination of the creator’s experience and insight up to that point.

A mathematical approximation of what a blog post might look like does not have such meaning. It is a mere facade, an illusion of a meaningful piece of content.

Navigation

So after all of that, you may find it hard to believe that I’m not just here to rant about social issues. In fact, my goal here is to actually tie this back to those of us working with technology, and find solutions. There are ways to help alleviate these issues, and work towards the good of society with the tools that we build.

In fact, it’s us working in the technology sector that have the most direct impact on these issues. While obviously the loneliness epidemic and meaning crisis both have deep roots, technological applications are a significant influence. However, there is a delicate balance to strike. New technologies are powerful, and have potential for real beneficial usage. Finding this balance is key, and hard.

Social contagion

First, navigating contagious meaninglessness. On a personal level, I struggle with what’s been referred to as “doom-scrolling”. Unfortunately, Instagram Reels and YouTube shorts have made up a sizable percentage of my focus over the past few years. This is time that I have simply wasted, and worse, I’ve spread it to friends of mine. When I see something adequately entertaining, I send it to those I think would agree.

This sharing of content to those within my network is actually the first step of the engagement trap. When a friend of mine sends me a batch of reels, I’m obliged to go and check them out. This gets me onto the application, and increases the likelihood of me spending time there. This is the “contagion” that I alluded to earlier, or at least part of it. Social media is social by nature, and so in order to combat the profit-driven attention traps, working together with friends is a positive step.

On a mental level, simply being conscious of the value that a given piece of content is providing to you is crucial. Short-form content is entertaining, for sure, but how many of these minute-long videos do you actually remember? For me, there are a few. Very few. That’s not to say there’s no benefit to short-form content, recipes are great in this format. Simply understand what you’re looking at and how it aligns with your long term goals.

Substack as a case study

On that note, for us builders, there are some examples of applications which have more positive social influence. For example, Substack. Substack doesn’t buy into the same engagement-to-profit pipeline that has become popular. That’s because the business model of Substack is fundamentally different.

I think I may do a more comprehensive case study on Substack from a technical, business, and design perspective. Let me know if that’s something you’d be interested in.

Substack’s model is to provide a practical platform and network for writers to utilize in order to share and monetize their content. That means that the content, which succeeds, is that which provides value. This also means that there isn’t an incentive to keep engagement, rather the incentive is to provide users–both readers and writers–with avenues for connection.

Writers are incentivized to create good content in order to attract readers who will find it valuable. Readers are then incentivized to pay in order to obtain access to exclusive perks. The business model does not promote meaninglessness, in fact quite the opposite. If Substack can maintain this, then looking into the future, this platform will likely become a treasure trove in a sea of garbage.

Striking the balance

Language models are no-doubt one of the coolest things that humanity has developed in the recent past. Throughout this article, I’ve detailed some of the ways that they bring us ever closer to a cyber dystopia. However, I’m someone who uses LLMs daily, and integrating them into applications has been a favorite project of mine. I have some thoughts on how to do it ethically, while maintaining meaning.

First, there are some things that LLMs are excellent at, and some areas where they perform poorly. I should really just make this point its own article at this point, so I won’t elaborate too far here. However, the most important lesson that I’ve learned is that language models work much better working off of a point of reference than working from scratch. They should be complimentary, not replacements.

For example, with regard to content creation. I’ll oftentimes use ChatGPT when writing my blog posts. That doesn’t mean that I say, “Hey ChatGPT, write me a blog post based on these key points.” Rather, it means specialized review of my content’s structure, grammar, voice, etc. ChatGPT can also be quite useful for research, and ideation.

The point is, it’s only improving my existing work, not trying to create it from scratch. Using LLMs to compliment human creation is the way that we use them effectively, in my opinion. Approximated language isn’t very useful all by itself. There’s a ratio to maintain between the human and LLM’s influence on a given creation, and it’s probably best to maintain a strong bias towards the former.

Like an onion

Especially regarding communication, it’s important to remember that there are layers of translation between you and your recipient. The more layers to that communication, and the further you place yourself from the other person. Beware of exacerbating the disconnection. Maybe just send the email you want to send. Use some professional language, but make it feel a little bit more human.

As we move towards metal, the warmth of the human touch becomes precious. Don’t lose that touch. Build your products to preserve that touch. Enable true human connection, allowing for more complexity and nuance to make its way through your channels. Convey emotions, parse expressions, whatever it may be. Build a direct bridge between people, don’t push them further apart.

That’s not to say that there can’t be some neat AI features helping you communicate. However, these features should actually help people connect better, instead of putting up a facade.

To learn more about how to successfully integrate AI features, check out my article on Invisible AI.

Build for the future we want

What it comes down to is this:

As technologists, we have the ability and responsibility to shape the tools which act as society’s infrastructure.

In the age of language models and whatnot, meaninglessness abounds. Loneliness intensifies, connections die. But the campfire isn’t out yet. We’re still the same humans, and we’re probably not going anywhere. So have hope, and just build the future that you want to see.

I can’t say exactly what that future should look like, although I have some opinions on how it shouldn’t. I’d like to think that we have the ability to create a positive environment for future generations.

Author’s Note

So this was kinda weird for me to write. I felt a lot of passion go into it, but I also have never published anything like this online before. I really hope you liked it, though, or at least found it useful. Furthermore, I think that we’re in a very strange time and there’s always this strange dystopian feel to the future. The uncanny valley is almost a tourist attraction at this point, with how often it’s visited.

But as always, thank you so much for reading. Your interest is actually so meaningful to me. I know that there probably aren’t a lot of you reading this, but to those who are, let me just say it again: thank you, and I’ll see you next week! Goodbye.