What AI means for engineers and creatives

I wrote all the text that you read on my personal website. But I didn’t write all my own code.

Following the monumental rise of AI usage in nearly every person’s daily workflow, deciding where and when its usage is appropriate has been one of the most pressing questions of my personal growth and identity. I consider myself both a creative, someone who pours time and effort into artwork representative of himself and his values, and an engineer, someone who works hard to efficiently and effectively produce something useful to himself and to society. Artificial intelligence simultaneously threatens and strengthens these respective pursuits. Distinguishing where AI is useful and harmful has been essential to reconciling this new, powerful technology with the things I love to do.

For creatives

Many artists regard the rise of generative artificial intelligence, for some its mere existence, as abhorrent. In the way it has begun displacing the essential human aspect of art, music, film, essay, even conversation, it is hard not to hate it. The word slop has been employed as a vivid descriptor of the quality of the things AI produces. And in many cases people find a “worse” piece of art more admirable or acceptable solely for the fact that it was made by a human rather than a computer.

As a musician and author, I align myself with all of these sentiments. I do not make music or write words to produce something “good” so much as I want to express myself and my identity through the fabrication of art. My songs and stories are an extension of myself, and so it is very important that they are wholly intellectually owned by me. It is true that my sense of harmony and melody, my knowledge of music theory, my vocabulary, my use of imagery, and so on are influenced entirely by the work of other humans, in the same way the training set of all AI models is built on human work. But the specific art I have consumed to influence the way I compose and I write is also a part of my identity. So it’s impossible to equivocate a song composed by me, or any other human, to a song composed by AI. While one is influenced by a person’s musical background and personal experiences, something produced by AI is founded only on a mathematical formulation of everything it has ever seen, which is most publicly available music at this point. One is unique and part of a person’s individual identity while the other is as unspecial as art can get by the nature of how it is produced.

Proponents of AI art suggest that AI is an accessibility tool for individuals with low artistic talent. Others suggest that the quality of one’s prompt plays into the quality of AI art. I find both of these arguments disingenuous because they mischaracterize what art is and what it is for. Art is a component of human culture. When people connect over art, they connect over the cultural ideas and artifacts represented in the art. The “culture” in question can be as wide as the human experience and as narrow as a single individual. But the important part is that a person, a member of the culture in question, materializes an idea into something tangible. Anyone can have an idea; they can have a style of song they want to create or a theme they want to embody in a story. But it is not the idea that is special; if it were then a story would be a meaningless meander toward its moral or lesson. But we know that a story is more than gratuitous fluff around a meaningful core. We know that the art is in the progression of the story and the world it lets us inhabit. We know the art in a piece of music is not its climax but the buildup of the song’s motifs and its sonic layers that we get to enjoy in different ways each time we listen. A prompt is nothing. A prompt is not art, and the direction an algorithm takes the prompt is not either.

Let me make something clear: there is no barrier to entry for creating art, in broad terms of what it means to be an artist. Self-expression is unbounded, and it is not subject to any concept of perfection or ability. There is such a thing as technical proficiency, and society admires the “experts” of each discipline for their strength in their respective skills. Even so, that skill is something that must be earned through practice and natural talent; it is no concern of accessibility. The fact that any person can command a computer to make art that mimics Picasso does not confer Picasso’s talent onto any such person. That person has not accessed a higher level of talent. It’s worth noting that talent is not even measured in any objective terms, and that the value of art does not rest solely in its technical impressiveness. A young child’s Mother’s Day card can make for beautiful art as it is an authentic expression of the child’s identity in their love for their mother and an effective vehicle for the artwork’s message and values in its conveyance of familial love. It does not matter if any other person in the world could pick up a crayon and draw stick figures that rival the kid’s.

A child’s drawing is a somewhat inane example, but I believe it extends to adulthood as well. The point is, no mother should be moved if their child AI-generated their Mother’s Day card. Similarly I believe there is no pride or accomplishment in AI-generated artwork of any kind. This even applies more loosely to other forms of self-expression. Blog posts and essays should not be written with AI because they are also representations of what you know, believe, and feel. Even if you give a chatbot an outline of your key points and have it synthesize them into an essay, you lose control over the minutia of word choice and sentence structure. As said before, expression is more than just the idea! It’s also the execution.

For engineers

Anyone who claims that AI is not a good programmer, not a good engineer, not a good mathematician, not a good writer, not a good biologist, et cetera, is kidding themselves. AI is frighteningly talented at most human tasks when exercised with careful control. It is notorious for producing identifiable patterns in its work, especially in writing, but these manifest primarily in lazy prompts which relinquish control of most of the task to the chatbot. In my experience, the quality of a model’s output scales dramatically with the degree of detail the user is willing to put in.

Since I began to learn programming, I have always been proud of being good at what I do. I imagine many others feel threatened or scared by AI in many ways becoming better, as if the human role in engineering is evaporating before our eyes. But I believe there are many reasons to be optimistic about AI’s impact on everyone’s work. While the floor for the quality and efficiency of one person’s work has been drastically increased, so has the ceiling. Anybody can vibe-code an app in an afternoon, and many people do. But this oversaturates the marketplace with half-baked products that no one would pay for because, well, they too could produce it in an afternoon! The value of a product still scales with the amount of human effort put in—many people working for many weeks on the product, with the help of AI, will produce something that far exceeds the work of one person in a day with the same help. While it’s true that both parties best any individual’s work a decade ago, the role of the human is not obsolete.

There are tradeoffs with regards to AI as well, though I believe many are mostly attributable to lazy practice rather than inherent problems with the technology itself. The effective skills of a project manager verifying the quality of their developers’ work transfers to the skills of a developer verifying the quality of their models’ output. Even if it is less efficient to produce code by hand, having a hands-on understanding of the inner workings of an app contributes greatly to one’s ability to maintain it, make architectural changes, pitch a product to customers, and fix issues, whether by themselves or with the help of AI. I have found it beneficial to engage with an AI agent as a codeveloper rather than a subordinate.

The advantages for using AI are even clearer outside the contentious world of agentic coding. Using AI to answer queries related to programming and to learn about programming in general is far superior to older methods of reading documentation or using online forums. AI has distilled all the useful information from these sources into an intelligent, conversational tutor on any subject. (And AI won’t rudely chide you for your ignorance!)

I’m still someone who enjoys paging through documentation and reading programming language specifications for fun, but this is because I find comprehensive expertise in a subject to be intrinsically rewarding, and because I consider programming to be an art form depending on the project at hand. There are scenarios where I use no AI whatsoever in my programming, because I believe code, like music and words, can be beautiful or it can be sloppy, and it can be personal or it can be impersonal. For projects that are personally important, that is to say part of my identity, my use of AI becomes far more moderated. You’ll find me asking it questions about programming rather than asking it to write my code for me in cases like those.

Ethical and logistical concerns

As AI is such a new technology, we have yet to fully understand its tradeoffs and consequences or enjoy using it in an equilibrious state. For one, there are environmental and energy concerns. AI is expensive largely because of its energy consumption, but because AI companies are highly subsidized by venture capitalist investments, they can provide their services to users for free or at a much lower cost than what is realistic for the resources being used.

AI currently falls into the category of other amenities I enjoy that are also harmful to the environment or are otherwise unethical but are also too personally or generally useful to completely abandon. Among these are the use of plastic packaging, taking flights, and using the internet in general. The dilemma is that, just as flying in an airplane is an important aspect of travel, AI is hard to ignore as a dominant part of every modern software developer’s workflow.

There’s an optimistic part of me that believes AI can “pay for itself” in that regard; that like other technological advancements it has the potential to do net good for the world even if some of its immediate side effects are harmful. On an individual scale, I feel that by making me a more efficient engineer AI also enables me to more efficiently do good—and in that perhaps offset the ways in which AI use harms the world. The baseline standard I hold myself to is that AI is a technology only worth using responsibly. If that is to the end of efficiently and effectively producing something useful to myself and to society, I am able to put some of my ethical qualms aside—especially as AI is used prominently for so many other nefarious things, like disinformation and cyberattacks. It’s tempting to say it is the technology itself that is evil and valid to advocate for legislation regarding its use, but as in all things, each individual is responsible for their own actions and cannot offset blame to the instruments of their misdeeds. At the same time, there are a lot of questions surrounding the ethical use of AI. I take ownership of my frequent use of AI, and I hope its benefits offset its cost. If that is wrong, that is my own moral failure, and not something I get to blame technology for. Every day I continue weighing the advantages and disadvantages of this new and powerful tool and the implications of its use.

That’s all to say, AI plays a complex and important role in my life. I like to be transparent about my perspectives and how I use it, and I think it’s important for every creative and every engineer to do the same.