“With the release of ChatGPT, I have started feeling more secure about the career choices I’ve made and the increasing importance of the humanities. I now realize my skill set will be one of the last to be replaced by a robot.”
The newly unveiled GPT-4 failed its AP English Literature and Composition exam. It passed every other Advanced Placement test with flying colors. AP English Lit is a test designed for high school students, completely rudimentary compared to the work of a PhD or Masters graduate.
In the humanities, there is no correct answer, only subjective perspectives that often bring up more questions. ChatGPT has already passed the LSAT and the Bar exam–which are still tests where there is one correct answer, regardless of whatever process of interpretation might lead to it. Unlike writing poetry.
In areas where emotional intelligence is necessary, AI will continue to perform poorly. It cannot innovate. It lacks the empathy to form a unique interpretation or empathize with the characters in a work of art. Despite the fact that it knows Art History better than your average college student.
“I will really start to worry when OpenAI releases a model that can write a good novel. I think at that point very few jobs will be safe.”
Though we are likely at least several decades away from such a development. If our experiments with self-driving cars have taught us anything, it’s that the last 10% of any technology is what takes the most time to develop.
Though I’ve always had an aptitude for technology, (one that is practically genetic considering my father’s line of work), I also have a love for literature. I could’ve studied computer science in college, and at the risk of sounding arrogant, I think that eventually I could have excelled at programming.
It would’ve been a more practical choice, I could’ve become a software engineer and entered a more lucrative career with opportunities for growth and a healthy work life balance. I disregarded that in favor of a field where the objective is to form a unique perspective, where there is no objectively correct answer. I was a technology geek but novels and poetry expressed something that seemed more important and interesting than rationality or logic. So for undergrad, I majored in English Literature .
Three years later, I struggled to find a new job, and one of my closest friends was also mourning his college major in psychology, a financially questionable field without extensive graduate school. He and I decided to sign up for a coding bootcamp. At the last minute, I was accepted to a creative writing program and decided to work on my novel instead. He went onto a successful career in software engineering and cryptocurrency. The University of Nevada Las Vegas is a selective and fully funded program headed by some of the most brilliant minds in creative writing, and I got to spend the last three years in the equivalent of a writing bootcamp. I was immersed in words and stories, exploring narratives in fiction and poetry and teaching undergraduates how to write.
I was never the best professor, and after I graduated I lamented the fact that teaching was the career path I had the most training in. I conveniently forgot about the personal and emotional developments I had made in the program, the relationships I nurtured and the writing skills I honed. This all seemed irrelevant as I entered my thirties and finally realized the importance of financial freedom. Ironically, ChatGPT has me feeling less pessimistic rather than terrified.
For years, the conventional wisdom was to learn how to code or how to analyze data, careers that were supposedly future-proof. New developments in AI contradict this theory. These are fields where there are objectively correct answers. Large language models like ChatGPT already excel at solving these kinds of problems.
Even though we only supposedly have access to AI technology which is already obsolete. ChatGPT can already replace a junior developer or data scientist. Of course, highly skilled experts in these fields are unlikely to be replaced anytime soon.
“ChatGPT still thinks all poems rhyme, no matter how many times I tried to teach it about free verse.”
It seems the model cannot make a choice that does not rely on a rationalist perspective, it cannot generate language that is random or nonsensical. Even if examples of this already exist online. It’s a poor post-modernist. Even though Walt Whitman’s poems are public domain and published all over the Internet, the model still can’t understand that his primary contribution to poetry was the move to free verse. Free verse does not follow a set pattern of instructions. Unlike math or high-school economics.
While SEO mills and sponsored content farms like CNET are already using AI to write their articles, higher quality publications that rely more on skill rather than repeating keywords will only ever use AI for research or basic outlining purposes.
Even a time where a student can learn almost anything from a Youtube video, the guidance of a human teacher is still necessary. As is the emotional intelligence of a psychologist, even in a world where diagnostic tests are replaced by computer overlords. Humanity will still be safe and well, as long as they don’t replace my bartender.
