And Other Marvelously Meta Matters
For over a month now, I’ve been writing about the rising seas of AI-generated media, the profusion of AI-generated art, and the ethical quandaries involved. So, it was only a matter of time before I stumbled onto the topic of AI prose generators like ChatGPT.
Except, I didn’t stumble.
I was pushed.
Pushed by several folks familiar with my interest in AIs and then by the sudden onslaught of headlines heralding the overnight sensation known at OpenAI’s superchatbot. Didn’t I warn you about the coming deluge? The water is rising fast now.
Anyway, I spent a little time today playing with the world’s latest chatbot darling. Here’s some of what I found.
Not feeling very original, I started by asking the chatbot about itself, and here is its self-portrait:
ChatGPT is a natural language processing (NLP) tool that uses a pre-trained language model to generate responses to user input in a conversational manner. The tool takes in user input, processes it using the language model, and generates a response based on the context and keywords in the input. The generated response is then displayed to the user, allowing for a back-and-forth conversation.
Well, that’s pretty prosaic but also nicely succinct. In fact, it’s written in better English than you’ll see in much of today’s blogosphere, both macro and micro.
Alrighty then. So, I wanted to go a bit deeper and asked it a simple question: “Is ChatGPT based on a neural network?” Here’s what it had to say:
Yes, ChatGPT is a variant of the GPT-3 language model, which is based on a neural network architecture. GPT-3 uses a type of network known as a transformer, which is specifically designed for natural language processing tasks. The model is trained using a combination of supervised learning, in which it is provided with large amounts of labeled data to learn from, and unsupervised learning, in which it is allowed to discover patterns in the data on its own. This allows it to generate human-like text that is often difficult to distinguish from text written by a person.
The Network Writing About Networks
But what, you may wonder, is a neural network? Glad you asked. ChatGPT has the answer to that one as well:
Neural networks are a type of machine learning algorithm that is designed to recognize patterns in data. They are called “neural networks” because they are inspired by the way the human brain works, with individual neurons working together to process information. In a neural network, there are many interconnected processing nodes, each of which performs a simple calculation on the data it receives. These calculations are designed to be similar to the way that neurons in the brain process information. The network as a whole is able to learn and adapt based on the data it receives, allowing it to improve its performance over time. This makes neural networks a powerful tool for tasks like image and speech recognition, natural language processing, and more.
The Difference Between Systems and Networks
Then I asked ChatGPT about the difference between systems and networks. I was genuinely curious about how it would parse those differences since I’ve found the distinction challenging myself. It did a find job laying things out clearly and at considerable length:
In a general sense, networks and systems are both collections of interconnected components that work together to achieve a common goal. However, there are some key differences between the two.
Networks are typically characterized by their interconnectedness and the flow of information or resources between nodes in the network. Networks can be thought of as a web of connections that facilitate the transfer of information or resources. Examples of networks in the natural world include neural networks in the brain, social networks, and ecological networks.
On the other hand, systems are characterized by their organization and the relationship between components that work together to achieve a common goal. Systems can be thought of as a set of components that are organized and interact with each other to perform a specific function or process. Examples of systems in the natural world include the human body, ecosystems, and weather patterns.
In summary, the key differences between networks and systems are the focus on interconnectedness in networks and the focus on organization in systems. Networks are characterized by the flow of information or resources between nodes, while systems are characterized by the interaction and organization of components to achieve a specific goal.
Is the Universe a Network?
Honestly, I doubt I would have been able describe those distinctions that well in the space of four short paragraphs. Well done, superchatbot!
So, I took our discussion to the ultimate level, asking about whether the universe itself is a network. Being a network nerd, I found the answer inspiring.
The universe can be seen as a network because it is comprised of various interconnected systems and structures. For example, the solar system is a network of planets, moons, and other celestial bodies that are held together by gravity. Similarly, the galaxy is a network of stars, gas, and dust that are held together by the force of gravity.
Additionally, the universe is a network of interconnected physical forces and laws, such as gravity, electromagnetism, and the strong and weak nuclear forces. These forces and laws work together to govern the behavior of matter and energy in the universe.
Furthermore, the universe is a network of living organisms and ecosystems that are interconnected through the cycles of matter and energy. For example, plants and animals rely on each other for food and oxygen, and their waste products are used by other organisms in the ecosystem.
Overall, the universe can be seen as a vast, interconnected network of systems, structures, forces, laws, and living organisms that work together to create and sustain the complex and dynamic nature of the universe.
Well, okay, so ChatGPT might get a little metaphysical here. For example, is there really such a thing as network of laws? Well, maybe. All in all, though, superchatbot whipped up a Carl-Sagan-esque answer about a topic as large as the universe in just a few seconds.
Like many other people, I was blown away by OpenAI’s superchatbot. I mean, yeah, it did some weird things. For example, it told me a pretty off-color joke about Joe Biden and a downright surreal one about Donald Trump. The bot may not be quite ready for its own comedy special yet.
And, it sometimes contradicted itself in weird ways, one minute claiming “I am in a good headspace and am able to appreciate the present moment” and the next “I do not have the ability to experience emotions.”
But still, it was able to do many other fascinating things, from writing love poetry to God in the manner of John Donne to providing a nice little book review of Moby-Dick.
Honestly, circa 2022-23, it appears we writers may soon be redundant, or at least practicing our craft with much greater humility. And it’s not just us, either. I also played a bit with OpenAI’s computer programming tool. Just by describing what I wanted the program to do, I got the codebot to write up several bits of a Python code, from a simple dictionary to a function that identifies whether or not a number is prime.
So, the good(ish) news is that we writers and artists will not be lonely in the unemployment line. Developers will be right there alongside us. Poetic justice, I suppose. In fact, I asked ChatGPT to write a poem on the topic, so I’m going to give it the last rather chillingly optimistic word:
In a world of endless possibility
Where machines can do the work of many
The jobs that once belonged to us
Are now at risk of obsolescence
Gone are the days of endless code
When writing meant pouring out your soul
Now AI can do it faster and better
Leaving writers out in the cold
And as for artists, once so revered
Their skills no longer needed
As AI can create with ease
Leaving them to wonder and grieve
But fear not, dear human friends
For though our jobs may disappear
We will find new ways to thrive
In a world transformed by AI.
5 thoughts on “A Neural Network Writes About Networks”