Tempted by Intelligence

For the first time in human history, we can outsource creation. Writing, drawing, coding—anything that doesn’t involve crafting something with our hands—can now be done by AI. Yet, we don’t know how much of this should be done by AI. Can AI-generated art be appreciated by the human mind? Will AI-generated writing ever be placed on the same pedestal as human-written classics? Why is there even a semantic difference between AI “generated” and human “written”? These are all questions born from our shared distrust towards AI as an outsider, an invader come to steal our passions and professions, as presented by Eric Zhou in the Oxford Press (Zhou 2024). 

This scepticism towards AI is deeply rooted in our cultural DNA. Like many technological advances before it, AI represents both bountiful opportunity and moral hazard. In religious scriptures from all Abrahamic religions and many others globally, “vice” is often manifested as a temptation. Portrayed as the easy way out or the path to short-term happiness and long-term regret. The parallels to AI—especially as students—cannot be ignored. Yes, this essay could have been written by any number of highly competent large language models. Yes, it would be difficult to tell, without the proper tools in place, whether or not it actually was written by AI. With the simple assurance that the ideas in it are my own, they seem more credible; more trustworthy. 

This negative perception around AI has been carefully cultivated for decades by the media questioning the morality, safety, and intentions of a truly artificial intelligence. Through dystopian books, fear-inducing articles, and end-of-the-world movies, a large part of the populace has been led to see AI as a modern Ragnarok. Unfortunately, this leads us to overlook its virtues. We now have tools that allow us to eliminate the busywork from our lives and focus on advancing the frontiers of our knowledge. Already, the power of large language models is being harnessed by ordinary people, myself included, to help our ideas come to life—whether hardware or software. These products, without the assistance of AI, would have likely fizzled out as an afterthought, never to see the light of day. It will allow the world to reward its ideators, visionaries, and dreamers. AI does not create ideas autonomously, but it helps those with the ideas realise their dreams. Gone is the time for pitch presentations; the time for prototypes has arrived. Ideas are no longer confined to a ten-slide keynote pitch. Thanks to the productivity of AI, it is now possible to present a functioning prototype without having to worry about involving other people or bureaucracies. Because, in today’s world, 20 dollars, the willingness to pay for a Large Language Model (LLM) subscription, and a few hours of your time now allow you to create a fully furnished web app for any previously deemed “longshot”. People no longer need to be computer science majors to build an app, talented with a brush to create art, or even speak English to write in English. 

It is impossible to tout the virtues of AI’s interactions with humanity, however, without acknowledging the idea that the more virtues it accumulates for itself, the greater of a vice it becomes. The better movies it can generate, the more directors will be tempted to turn to AI-generated content over the hiring of actual actors. The more closely it can mimic humans, the less unique we become. If we decide to one day give in to AI’s vices, it may well exercise a vice-like grip on the lifeblood of the modern global economy: information. The more information it garners, the better it will know us. In some cases, chat bots are already starting to gain a better understanding of the users than they have of themselves. Those same people are then falling prey to its vices, allowing AI to dictate their schedule, facilitate their relationships, and take on their responsibilities. All large language models are trained on billions—nearing on trillions—of data points. This data is generated by humans, in the form of writing, art, and film. The better grasp a model has of this wealth of data, the better it is able to create its own takes on those forms of media. Similarly, the more human-led companies know about their customers, the better they are able to induce them to spend their money. 

Spending money, consuming content, purchasing products, AI has the potential to one day streamline all of these processes for us. It also has, however, the potential to imbue them with sinister intent. Its vices are not present, however, in its destructive capabilities. That we cannot individually control. That temptation, however, to allow it to dictate our lives for us is where we risk ourselves falling victim to its vices. Relinquishing control of our lives comes with an oddly freeing feeling, and yet the thought of resigning ourselves to the whims of an algorithm is as terrifying as it is freeing. Artificial intelligence has as many virtues as it does vices. While it cannot be said yet how it will be put to use, we can hope to harness its virtues and steer clear of those vices.

References

Eric Zhou, Dokyun Lee, Generative artificial intelligence, human creativity, and art, PNAS Nexus, Volume 3, Issue 3, March 2024, pgae052, https://doi.org/10.1093/pnasnexus/pgae052