Recently, AI language processing technology has taken the world by storm. ChatGPT is a type of AI that can help us with all kinds of tasks from content production to coding.
However, as anyone who has used the platform can attest, there are clear limitations to what ChatGPT can do.
So how much of the recent hubbub indicates an actual revolution in the text generation space? Is this AI tool a true technological revolution? Or can we chalk most of it up to hype?
This article will explore how GPT works, its limitations, and how this nifty piece of technology will actually influence our lives.
What is ChatGPT and how does it work?
“ChatGPT is the latest in artificial intelligence technology, utilizing a “generative pre-trained transformer” to create natural language conversations. ChatGPT works by applying advanced machine learning algorithms to break down distinct elements of language and then enhancing them with an in-depth understanding of contextual interactions.
ChatGPT’s intelligent software can generate responses that mimic human conversation, providing interactive dialogue that is surprisingly natural and fluid.”
The above-italicized text was generated completely by AI with no human edits. As you can see, it shares some informative facts about the topic in a format that resembles human conversation.
A rough overview on how the AI is able to do this, is that is takes information from huge dataset and uses machine learning to combine the info into complete sentences.
ChatGPT specifically use text databases that were pulled from the internet between 2016 and 2021. Machine learning parameters combine the information into different combinations of text to create the end product.
This truly is a nifty trick where you can seemingly “ask” the tool a question, and a coherent “answer” comes out the other end.
But what is this “answer” that you are getting really? It is coherent text sure, However, the usefulness and impact of this nifty trick may be overstated when you look deeper into its limitations.
What are the limitations of ChatGPT technology?
While ChatGPT certainly has the capability to spit out some coherent and useful text, it has some glaring limitations:
One of the most obvious limitations of AI technology is its reliability. The function of using pre-trained text to generate an “answer” does not have a built-in fact checker. Just because something was said before (somewhere on the internet between 2016 and 2021) doesn’t mean it is factually correct.
ChatGPT does get things wrong sometimes. When you consider the implications of this limitation, it invalidates the tool as a useful source of truth. This is because even if the tool is right 99% of the time, there is no way of knowing whether the fact it confidently stated is the 1% time that it is wrong.
Until ChatGPT can be right 100% of the time, users will always be forced to fact-check on their own to be sure they are getting the right information.
ChatGPT is “pretrained” with data from before 2022. This means that it can’t generate responses about current events. As it turns out, the most recent “breaking” information is often the most valuable to people.
Therefore, until AI technology can find a way to be completely up-to-date, people will always search elsewhere for recent information.
3. Reasoning or Critical Thinking
As humans, the main function of our thinking is to reason. Reason allows us to align our thoughts into useful patterns and make decisions.
Right now, ChatGPT can mimic reasoning but doesn’t actually have any decision-making capability at all. Part of the trick that is performed by the generative language model is its apparent confidence in the answers it provides.
This confidence is not derived from the critical thinking faculties that humans are capable of. It is simply regurgitating already reasoned points that were stated somewhere in the dataset from which it crawled.
In my mind, a lack of reasoning capability is a true dagger in the heart of the AI evangelist’s pet project. This is because calling ChatGPT a true “Artificial Intelligence” is simply incorrect.
The fact is that no matter how complicated and advanced this technology becomes, it will still remain incapable of reasoning or critical thinking.
4. Conveying Meaning
To a robot, it may be hard to decipher the purpose of flapping our jowls at each other, scribbling lines on a notepad, or clacking away on a keyboard. For humans, the primary function of writing, speech or text is to convey meaning.
Language evolved over the millennia as a method to share insights with one another. These insights allowed us to combine valuable knowledge and wisdom to work together toward common goals.
ChatGPT is incapable of sharing “valuable” insights with humans. This is because it is not in a position to understand what is valuable.
Sure, the text bot can rehash existing statements that contain meaning within them. But the value of those statements is highly limited in their contextual relevance. This is because generative language processing cannot infer meaning in the questions it is being asked, or produce meaning independent of the inputs it is provided.
While ChatGPT can seemingly adopt moral stances on certain issues, it is not capable of expressing true morality. This is because moral behavior has to occur in antithesis to immoral behavior.
As a pre-programmed machine, ChatGPT cannot actually make a moral choice because, well, it doesn’t have a choice. The more complicated the technology becomes, the more convincing its moral stance may appear. However, by definition, a preprogrammed machine does not “choose” anything. Rather, it performs a computation based on its inputs and spits out an answer.
In contrast, humans make choices constantly. Those choices have repercussions and therefore have moral implications.
Note: The determinists may jump down my throat for this argument and fair enough. So without getting into a deeper conversation about determinism vs. free will, I ask you to concede that humans at least believe and behave like we have free will while the chatbot does not.
How will AI technology affect our lives given its limitations?
The ChatGPT AI tool certainly has some utility in the modern world. However, I am not convinced that it will spur the next technological revolution that many imagine.
Given its limitations, I am convinced that ChatGPT is a fancy text tool that can help humans get their ideas out.
Maybe some of the more rudimentary limitations such as reliability and recency can be incorporated into the current model over time. But these upgrades will still not address the fundamental limitations which hold it back from becoming a TRUE AI – reason, meaning, and morals prove to be monumental hurdles for the next iterations.
I anticipate that many people will lean heavily on this technology now that the cat is out of the bag (especially people in the content production space). New business ventures and opportunities will be born, capabilities will be pushed, and we will quickly become accustomed to a world with generative language transformation.
But ultimately, people will become bored with it.
If the purpose of language is to convey meaning, then after a while, the tired rehashed meaning produced from a chatbot just won’t cut it.
As it turns out, text in and of itself is not useful or meaningful – even if it’s arranged in an entertaining or useful way.
Meaning requires context and perspective. Without them, text is nothing more than arbitrary symbols.
To possess meaningful perspective requires human insight – otherwise, text has no context. Therefore, humans will always be needed in the content production process no matter how complex this AI technology becomes.
Don’t believe me? Ask chatGPT how the limitations of AI will affect our lives. Read its output, and ask yourself “did I gain any meaningful insight into this question beyond the obvious?”
If the answer is yes, great, disregard this post because it would seem that the technological revolution is neigh – congrats, you no longer have to think for yourself.
If the answer is no, case in point. Life goes on.
ChatGPT is a neat tool. There is certainly a wow factor associated with a machine that can interpret text input and produce coherent text output.
But let’s not fool ourselves. This technology is a long way off from revolutionizing the way we share and interpret information. Given the current limitations presented in this article, we will not see machines take over the world terminator style in the foreseeable future.
Many jobs will be impacted by this technology – copywriters, content producers, editors, and SEOs to name a few. But these jobs will not be replaced by robots any time soon.
Maybe one day a true AI will be produced – one that is capable of reasoning. That day will likely usher in a new age. Until then, we humans will have to rely on our decision-making faculties to build businesses and produce value in the world.
Luckily, if you’re interested in building a business, you’re in the right place. This website is dedicated to teaching you how to do that. Stick around to learn more about how to make money online the real and sustainable way.