Luka Misic

13 March 2023

Since the earliest days of humanity, technological advances have been transforming society. In the time of the hunter gatherers, the agricultural revolution led to a fundamental change in the way we live our lives. With the procurement of a stable food supply, humans were able to settle into sedentary societies, and the population began to explode. Now that humans could focus on something more than their day-to-day survival, they were able to turn their minds to more abstract concepts. Around the globe, the first true human civilizations began to flourish.

Many centuries later, humanity would experience another foundational shift. The technologies of the industrial revolution intruded into every aspect of our lives and societies. The schedules of the early factories created what would later become the hours of our modern workdays. Advances in transportation such as the train or the steamboat linked our world together as never before. Communication technologies like the telegram and the telephone facilitated near instantaneous communication over previously unimaginable distances. Developments in weaponry set us on the course of the bloodiest and costliest wars in human history.

More recently came the emergence of the internet. Now, all of humanity’s information lay little more than a fingertip away. What once took days of toil to find could now be pulled up in little more than a second of work. Moreover, the internet would bind humans together from all over the world. Suddenly, a man from rural Wyoming and a woman from the heart of urban Paris could quite plausibly end up on the same information space together. The internet gave everyone a voice and allowed diverse and varied ideas to instantly disseminate over unimaginable distances. We were now bound together in a network of global interconnectivity, and the world was forever changed.

For years now, many people have speculated as to what the next technological revolution would be. Some have hypothesized that it would lie in the realm of space travel, allowing humanity to finally spread their reach to other worlds more than half a century after Neil Armstrong first hailed “one giant leap for mankind.” Others have suggested that this advancement would fall in the field of green energy, bringing about a breakthrough that would allow us to reverse global warming and to save our environment. Nothing, however, has captured human imagination more than the possibility of Artificial Intelligence: thinking machines that would have the capacity to carry on human functions.

AI has been the subject of intense discussion for decades. Some speculated that AI would lead us into a utopian post-scarcity society where humans no longer need to work, but can instead focus on the pleasure of the mind. Others, perhaps more frequently, however, forecasted futures of doom where humanity has been reduced to slaves by their own creations. Movies such as Terminator, Matrix, and WALL-E became staples of pop culture. Yet, as the decades tickled by, “Judgement Day” never came to pass. The Age of AI seemed something destined to remain forever consigned to the realm of science fiction and daydreams.

Now, it seems as if the moment of reckoning may finally have come. Over recent years, there has been a rapid rise in the use and creation of AI language models. The most recent of these AI, and perhaps the most notable, was ChatGPT, a chatbot launched by OpenAI in November of 2022. ChatGPT and its contemporaries have been designed to utilize deep learning techniques, training on massive amounts of text data to generate human-like responses to a wide range of questions. This has made them valuable tools for tasks such as conversation, translation, summarization, and more.

ChatGPT and other AI language models like GPT-3, BERT, and RoBERTa are programmed to learn patterns in language and generate responses based on that learning. They cannot be considered true AI, as they merely comport responses based on the amalgamation of data patterns and as of now, do not contain real sentience. Nonetheless, they are able to generate seemingly organic responses when questioned and are able to more than adequately serve a variety of functions.

Thus far, these advancements have allowed humanity to reap great benefits. But for all the good they brought the world, so too did they bring darkness. In the industrial age, for every phone or telegraph invented, we also gain unfathomable weapons of death and destruction. Because humanity did not adapt to take these new developments seriously as they arose, old tactics were able to clash with innovative weaponry to unleash heretofore untold horror and carnage in the First World War. Chemical weapons were outlawed afterwards, but a generation of young men would bear their scars until the end of their days.

We cannot afford to make such a mistake again. AI, more so perhaps than any of the other revolutionary tools that have transformed the world, has the potential to devastate the very structure of our society if its growth is left unchecked. Eliminating AI entirely is both impossible and undesirable in light of the boons it has brought and will continue to bring to the world. Nonetheless, we must introduce legislation to regulate the rate at which companies adopt AI technology to slow down the rate of change and give humanity a chance to adapt to a world rapidly shifting beneath its feet.

Already, we have seen AI begin to take on large roles in the corporate realm. Companies like Apple and Amazon were early adopters of the technology, using these language models to develop virtual assistants such as Siri and Alexa, which are capable of responding to user requests in a human-like manner. It was largely the novelty of these virtual assistants when they first launched that helped adopters gain a marketing edge over their AI-lacking peer competitors. 

Despite their relatively primitive nature, the actual functionality of these early AI language models cannot be understated. These AI models help users operate complicated functions of their devices through simple voice prompts. Additionally, they are capable of providing users with vocalized reminders when prompted and are even capable of carrying on basic conversations in their function as a form of chat bot. These factors all lead competitors to race to develop their own AI, triggering an evolving technological race pushing the envelope of the field forward to the present day, where we see the rise of more sophisticated models such as ChatGPT.

The impact of these AI language models has already been felt far beyond the basic auspices of the technological realm. Most notably, perhaps, has been their impact in the domain of education. Supplementing the traditional educational environment, AI language models have been used to create innovative new educational tools, such as personalized tutors and language learning applications. These tools have proven to be an invaluable tool to relieving beleaguered teachers in overlarge classrooms unable to devote equal attention to all their students. These AI tools assist teachers by responding to basic student questions and by helping them create customized learning experiences based on individual needs and abilities.

In the future, AI language models may lead to a full overhaul of the current educational system. AI language models will be able to effectively analyze a student’s previous performance along with their strengths and weaknesses in order to generate a customized study plan focusing on the areas where the student needs the most help. This could lead to a fundamental streamlining of the education process in which students would be directed towards areas where they need to learn, rather than wasting time on topics they have already mastered. This customized education system would solve long standing issues with previous education initiatives such as No Child Left Behind and Common Core. This system would strengthen the education of the weakest students, helping them keep up, without slowing down the progress of the stronger students alongside them.

While all these changes may seem positive on the surface, they also herald an insidious dark side. Free and easily accessible AI software like ChatGPT has heralded a new golden age of cheating to academia. On its own, ChatGPT is capable of writing surprisingly sophisticated essays that many high school teachers attest surpass the writing of their average student. When presented with the work of even a mediocre student, ChatGPT is able to refine it to create highly polished writing. As a result, many teachers have struggled to find new ways to assess their students’ writing ability. Others have gone so far as to question whether teaching writing is even an important skill anymore, or if it has been made obsolete in the same way the calculator has made practical knowledge of long division irrelevant.

In the graduate sphere, ChatGPT remains, much to teachers’ frustrations, effective. Recently, the chatbot was faced with a law test from the University of Minnesota and a business test from the Wharton School of the University of Pennsylvania. It passed both exams, albeit only at the level of a C+ student on the law test and at the level of a B to B- grade on the business test. As aforementioned, the chatbot is not likely to be used on its own in this manner. Rather, it would be used to supplement a student’s existing work or to provide the foundation of their work. Nonetheless, it stunts the development of research skills and is liable to provide unsuspecting students with inaccurate answers, at times. In the future, the usage of ChatGPT, as well as its AI contemporaries, will continue to grow more pervasive in academia as their programming further evolves.

The only way to potentially counteract this development is to introduce legislation forcing companies to release software capable of identifying AI written text alongside the release of their initial programs. This would largely do away with the most basic set of cheaters copying text directly from the software in lieu of putting any actual work into their assignments. While any software attempting to detect whether students are putting recombinant assignments together will be much harder to create, eliminating the copy-paste cheating will force students to put at least some effort into their assignments. Additionally, it can potentially be argued that the recombinant work of mixing one’s own ideas with that of the AI will be an essential skill in the future of our society and is not necessarily a bad thing in and of itself if applied in more appropriate contexts.

Overreliance on these AI to carry out basic functions also precipitates the development of dangerous workplace situations. If a lazy employee uses ChatGPT for instructions on how to do something at their job and it erroneously informs them, catastrophe may result, particularly if they work in a critical field such as healthcare or some form of vital infrastructure. The potential for societal catastrophe would only be compounded if this dependence becomes widespread. Furthermore, as with other essential technological systems, the program bears the risk of being hacked by malicious actors seeking to intentionally spread chaos or misinformation. As such, either the government, or the relevant institutional actors should take action to forbid the use of AI in these situations. While some will doubtless try to circumvent the restrictions, the presence of a barrier to use will greatly reduce the assorted risks.

These AI models will also have a calamitous effect on the job market. In education, we will soon see the creation of virtual teachers. This will mark a significant evolution of their existing uses. Rather than simply helping teachers structure their lesson plans, these AI virtual teachers would handle most educational functions themselves, answering students’ questions and providing the support necessary to succeed. While proponents argue that virtual teachers could be effective in areas where there is a shortage of qualified teachers as a cost-effective solution to filling the hiring gap, it is difficult to see how such a practice would not evolve to laying off qualified existing scholars. Many schools are run as a business first and an education model second. Even if AI is not eminently capable of matching the performance of skilled teachers, if it is capable of filling the role in some capacity, cost cutting schools will seek to take advantage.

In the field of customer service, where the technological demands are less advanced, we are already seeing AI language models begin to replace live human representatives. These AIs are able to answer any common questions a patron may have both quickly and efficiently. As previously stated, these AIs are cheap compared to any human hire. Even if the initial investment to create the AI is expensive, once operational, the AI requires no further financing. While additional investments may improve the AI and ease its flow of conversation, they are not necessary. In comparison, a human worker would need to be paid as long as they are on the company retainer and would likely need to be granted additional benefits such as health insurance or vacation days.

All this is to say that if a company is able to cut costs by replacing their human workers with comparable AIs, the laws of the market indicate that they are probable to do so. This will lead to a dangerous surge in unemployment with potentially societally destabilizing consequences as more and more people are crowded out of their jobs by automation and AI.

It is important to note that these developments are all happening in the shadow of the relative primitivity of AI technology thus far. Currently, AI is far from perfect and is still liable to make mistakes and give inaccurate responses when questioned. As a result, many companies are hesitant to use these AI models for much above the most basic of roles. Nonetheless, as the years pass, the AI models will continue to increase in sophistication and additional responsibility will continue to be handed off to them.

In the near future, AI will devastate job prospects in a number of fields. Artists are already seeing themselves supplanted by prolific AI generators. Any aspiring reporter must be aware that the days in which an AI can instantaneously generate articles on the day’s events are not far off. Even novelists and script writers are not safe. When given a basic prompt, AI will be able to perform months of arduous work in a matter of seconds. Any qualitative gap will be more than made up for by raw quantity. And this is not to say that the qualitative gap will exist indefinitely, either. AI is ever evolving, growing more and more sophisticated as we continue to add on to it.

To preserve these jobs, the government must take a firm stance outlawing the use of AI in certain industries, or at least placing an indefinite moratorium on their adoption until an alternative solution is developed. While critics will argue that it is absurd and immoral to attempt to slow the spread of technology, AI is societally disruptive in a far more fundamental way than most any previous advancement. When the horse was put out to pasture by the advent of the car, carriage drivers could easily transition to a new role as cab drivers. If AI is allowed to take primacy in the job market, there will be no such easy respite.

While the “Judgment Day” of the Terminator films may yet be some way off, the destabilizing risk AI poses to the foundation of our societies is very real and present. Humanity must find a way to cope with the threats posed by the development of AI and automation, or we will face an unemployment crisis in the near future. The warning signs are all flashing red. If humanity takes extraordinary action to address this challenge now, we may yet have a chance to avert the coming catastrophe. But if we delay, as we’ve done so many other times in our history, our fate is already set in stone. Any action that may yet come would be far too late to make a difference.

---

Image via Pexels Free Photos.

“Introducing Chatgpt.” Introducing ChatGPT. Accessed March 2, 2023. https://openai.com/blog/chatgpt. 

Ropek, Lucas. “New York City Schools Ban Chatgpt to Head off a Cheating Epidemic.” Gizmodo. Gizmodo, January 4, 2023. https://gizmodo.com/new-york-city-schools-chatgpt-ban-cheating-essay-openai-1849949384. 

Stokel-Walker, Chris. “CHATGPT Listed as Author on Research Papers: Many Scientists Disapprove.” Nature News. Nature Publishing Group, January 18, 2023. https://www.nature.com/articles/d41586-023-00107-z.

Herman, Daniel. “The End of High-School English.” The Atlantic. Atlantic Media Company, December 16, 2022. https://www.theatlantic.com/technology/archive/2022/12/openai-chatgpt-writing-high-school-english-essay/672412/. 

Bowman, Emma. “A New AI Chatbot Might Do Your Homework for You. but It’s Still Not an A+ Student.” NPR. NPR, December 19, 2022. https://www.npr.org/2022/12/19/1143912956/chatgpt-ai-chatbot-homework-academia. 

Rosalsky, Greg, and Emma Peaslee. “This 22-Year-Old Is Trying to Save Us from Chatgpt before It Changes Writing Forever.” NPR. NPR, January 17, 2023. https://www.npr.org/sections/money/2023/01/17/1149206188/this-22-year-old-is-trying-to-save-us-from-chatgpt-before-it-changes-writing-for. 

Kelly, Samantha Murphy. “CHATGPT Passes Exams from Law and Business Schools | CNN Business.” CNN. Cable News Network, January 26, 2023. https://www.cnn.com/2023/01/26/tech/chatgpt-passes-exams/index.html. 

“Chatgpt: Friend or Foe? – The Lancet Digital Health.” Accessed March 2, 2023. https://www.thelancet.com/journals/landig/article/PIIS2589-7500(23)00023-7/fulltext.