AI Isn’t Going Anywhere. We’re in an Age of Collaboration.
“What is artificial intelligence?” asked Quata reporter Ben Brubaker in one of the publication’s newsletters last year. “Depending on whom you ask, the term might refer to sci-fi entities, existing cutting-edge technology or even software we use every day.”
According to Brubaker, the term “artificial intelligence” was invented by computer scientist John McCarthy in 1955 to refer to machine behaviors that people would consider “intelligent” if these behaviors were observed in a human being.
But many things in the universe show intelligence. Writing in the 1970s, Mortimer Adler suggested that “it can be said that many of the higher animals think, and even that computers think.” As we will see, even the learning tactics of some AI closely resemble the learning patterns of people.
We live in a world where we encounter AI daily without even thinking about it. It corrects the misspellings in my texts and predicts the next word in the sentence of an email I write. AI isn’t going anywhere. It’s more than a geeky fad. It is here to stay, and if that’s the case, then we need to learn to work with this new technology.
That word “technology” is funny. Just like the blanket term “AI,” technology means different things to different people. Today, we think of the wheel and the fulcrum as rather primitive forms of technology. In comparison to a circuit board, the code on the backend of a website, a rover on Mars, or a fetal surgery, they do indeed seem antiquated. Nevertheless, they remain forms of technology. They helped get us to where we are today. It will be interesting to see where AI takes us, what sort of tomorrow will be forged with this new tool.
We’re living in an age of collaboration and adaptation. This is something which Shyam Sankar, CTO of the successful data and software company Palantir Technologies (this name is a pop culture reference to an orb in Lord of the Rings that allows characters to view things far away) is keenly aware of. In Sankar’s opinion, it’s not the place of AI to be a substitute for a real person, but AI can give the user an advantage.
If AI tools can give us a leg up on the competition, then it would behoove us to be neither scared nor indifferent of using it. AI does not replace the human workforce but helps it excel. These systems can shave hours off menial and monotonous work, allowing an employee to redirect their energies to more important tasks.
For the rest of this article, I’m going to focus on generative AI, and for a few reasons. First, because it is generative AI such as ChatGPT (and a few chatbots) that has captivated the public for the last two years. Second, because generative AI (especially LLMs) kicks out content at an overwhelming speed and, as a writer and communicator, this hits close to home.
LLMs, or large language models, are a specific type of generative AI that are trained on a vast quantity of data and produce coherent writing in response to a question or some other prompt. If we’re going to collaborate with AI, it is good to know what it is and a bit about how it works.
How Do LLMs Work?
I say “a bit” about how it works because I’m afraid that, when it comes to coding, I’m terribly ignorant. But short of that, I shall offer an interpretation of how LLMs work.
First, when speaking about AI, let’s use McCarthy’s concept — that of a machine’s behavior being similar to the intelligence displayed by a human being. That’s our definition. All AI is an attempt to replicate human behavior to some degree.
Under the broad label of AI, there are numerous systems that get trained in different ways and are developed for different functions. Narrow AI, for example, only performs very specific tasks; basic algorithms are an example of this. General AI, however, supports a wider range of abilities. Many of these systems rely on neural networks, machine learning, and recognizing patterns. The large language model (LLM) is a machine learning model. An LLM specifically studies the structure of a language and learns to produce human-like writing when prompted. But if it “learns,” that implies that it is taught.
This learning process resembles that of the formative years in a classical education. If an AI system such as an LLM uses a neural network, which is based somewhat on the human brain, then I’m not surprised to find it mimicking a child’s learning method.
In the classical model of education, before the rhetoric and logic stages (where higher thinking becomes crucial), the child goes through the grammar stage in elementary school. Here the child learns basic facts and notices the differences between one thing and another thing. In the grammar stage, a teacher pours a lot of information into the child’s mind: language, arithmetic, beauty, social cues, expectations, and more. The focus is on the input.
Something comparable occurs when training an LLM. The developer exposes the system to a dataset, often a pretty big one. This dataset, or collection of written text, is called a corpus. Like the teacher pouring letters and numbers into the child’s mind to become known, so too, the developer pours text into the language model to be assimilated. As it matures, the LLM is also able to learn from experiences in order to, hopefully, make better predictions or outputs in the future.
Pulling from its massive corpus (or corpse if we may be so crude), the LLM cobbles together some old parts to make a new whole, the result being not so repulsive as Frankenstein’s monster. In fact, most text produced by these generative AI apps is elegantly crisp and concise.
Are AI Applications Useful?
In short, yes — they can be useful. Across multiple use cases, Sankar sees saving time as a huge advantage over the competition. A real-world example he gave at the “America in the Age of AI” seminar from Washington Post Live was an insurance company automating their underwriting process with AI, saving their team hours of busy work.
Libraries have used machine learning (ML) for transcription and classification, thereby laying the groundwork for archival research in the future. AI could be implemented to help assess a student’s optimal learning style and even facilitate “more purposeful, efficient learning experiences.”
Analysis and content creation are areas in which AI might aid marketers and producers in any number of fields. Canva offers AI tools like Magic Write and Dream Lab to generate text and images. Other tools have a broader application. Adobe Acrobat’s AI Assistant can summarize (or synthesize) the content of a document, again saving time on analysis. Some tools can generate slides for a presentation — something that can eat up a morning. A slide-generation tool can create the desired product in seconds. And all I have to do is look over these slides to ensure the information is accurate.
If we find ourselves in an industry being reshaped by this new technology, and the reality is that most industries are, then we should be learning too. If AI is being trained to perform a certain task, then we should be trained on how to use this new tool to do these tasks. We should not fear being replaced by AI so long as we treat it as a tool and learn how to use it.
Like anything in life, here’s an area where, if we seek improvement, it’s going to take some effort. If I want stronger muscles, I have to go work out. If I want to be healthier, I have to cook better meals. If I want a stronger spiritual life, I need to pray more. If I want to do my job well, I need to regularly train, learn new skills, and become familiar with different software and interfaces. The same goes for AI. It is another tool to add to our repertoire, and it will take mastering.
We shouldn’t worry about becoming obsolete so long we try to stay abreast of developments in this technology and learn to collaborate with the tools it places at our disposal. It requires constant growing and implementing.
To become more familiar with AI, I can suggest a few digital learning opportunities I’ve come across. Common Sense Education offers an online course geared to teachers about AI foundations, and Brilliant.org offers a course on how LLMs work. And there are countless other options available.
Yet, as with any advancing technology, there will be misuse cases and drawbacks. In a classroom, for example, I can see where a chatbot might engage a student for a time and might even help teachers discern the method that is the best to help the student learn. But what the chatbot cannot do is replace the teacher. Alexander had Aristotle; Aquinas had Albert. A teacher is more than a speaker. The teacher must show, not just tell, and not only inform, but inspire. Passion and personal intrigue can only be passed on by a person.
Not only that, but a child knows when he or she is unwanted and being neglected. If a teacher uses a bot to babysit them, they will know it. And that is not good for their social and mental wellbeing.
It’s not just the sphere of education that should be of concern. The danger rests in how we interact with AI in healthy or unhealthy ways. Only the acting agent, a man or woman, makes the technology’s use good or evil. The wheel gives mobility to ambulance and tank alike. The fulcrum can be the base either for a child’s seesaw or a warrior’s catapult. So, while there are definite benefits from AI tools, they can also be poorly used. It’s up to us people of flesh and blood to choose how we use the technology sitting at our fingertips.
What’s the Difference Between Human and Artificial Intelligence?
Let’s again consider the analogy of education, how machine learning imitates human learning. It should be noted that one’s knowledge is only as good as one’s education, whether it be academic, formal, or personal. If a person is taught misinformation and biased history, then their own knowledge will be jaded, at least until someone sets them back on the straight and narrow path. Likewise, a language model is only as good as its corpus and training. If a prompt includes an unknown word (a word absent from the training corpus), an error occurs.
AI is dependent on human training or at least programming. Thus, while AI remains ever-evolving, it also remains indebted to its human designers, affected even by human flaws and fallacies. And what sets us apart, we humans from our machines?
The early Christian writers and philosophers expressed that the chief attribute which humans share with God is an intellect, or the ability to reason. As Augustine of Hippo wrote in his Confessions regarding God’s creation:
“…and we see humankind, made in your image and likeness, set over all these irrational living creatures in virtue of this same image and likeness to you, which resides in its reason and intelligence.”
But if artificial programs are intelligent, what makes us so special? Are we special? Well, we might say that these bots know more facts than us. But any teacher will tell you education is not just an assortment of facts. Machine learning models are programmed with logic, but there is more to human thinking than logic. There is imagination, something Albert Einstein greatly admired, and creativity.
The creativity and thinking displayed by humans is of a higher and truer nature than the statistical number-crunching and random variables put forth by AI. Mortimer Adler, following on Aristotle’s thinking, believes humans are unique in our identity as rational, or philosophical, animals. An AI does not question what truth is, what the essence of being might be, or what the purpose of life is.
An AI isn’t going to be able to have a conversation with an elected official and, in the course of an interview, ask the most important or interesting question, as a journalist might. A machine has no human experience. It won’t be able to pull from emotion or experience when it generates a poem. It is incapable of inventing altogether new words like the great authors such as Tolkien, Lewis, Dahl, and Seuss have. Good writers don’t just communicate well; they have to have something to say. LLMs don’t have anything, in particular, to say of their own accord. Their ramblings are the stitched-together newspaper clippings of the past and little more.
In an age where the ludicrous “Church of AI” exists and in which robo-preachers grace Buddhist temples, some may even question whether the vast intelligence of AI could be equivalent or superior to the divine. To this misguided assumption, I say simply that my God is the only one who creates ex nihilo, out of nothing. He needs neither corpus nor foundation, for He laid the foundations of all creation.
Will artificial intelligence achieve dominance over humanity? Could it destroy what we’ve worked so hard for? No, only humans could pull that off. But could it bring about a new creation instead? No, only God can do that.
