Skip to content ↓
MIT blogger Rona W. '21

taste by Rona W. '23

ai has no taste.

If you follow me on Instagram, Twitter, or (god forbid) LinkedIn, you might’ve seen me mention that I spoke to The Washington Post for an article about Gen Z and the job market.

But the explosion of AI has changed some young people’s paths. Rona Wang, who recently graduated from the Massachusetts Institute of Technology with degrees in math and computer science, turned down a tech job she thinks could be subject to automation. Instead, she opted to pursue a master’s degree in programming that’s closer to the hardware.

“Absolutely it’s about staying ahead of the curve,” she said. “A good rule of thumb is looking for jobs and skills that require [judgment] or research in some way.”

Zoomers aren’t ignoring possible harms, despite their excitement. Some say they’re worried about the implications of AI, including its ability to spread misinformation, make people lazy to learn, raise the bar for entry-level jobs and become a way for employers to cut costs — even if it means lowering the quality of work.

As you can see, the word [judgment] is in brackets, meaning it isn’t part of the original quote. From what I remember (please note my memory is faulty and not subject to the same journalistic standards as the WaPo), this is how the conversation actually went.

Me: “A good rule of thumb [in choosing jobs that will not become easily automated by AI] is looking for jobs and skills that require taste or research in some way.”

Danielle, the journalist: “What do you mean by taste?”

Me: “Like, having good judgment and discernment. It’s one thing to design a web app and another to build it, which can be very formulaic.”

I’m not sure if the quote that ended up in the article quite captures the same meaning as the one I intended—to me, judgment feels more mechanical, less human than taste—but of course this often happens in journalism and I appreciate the opportunity to get to speak on this topic anyway. (I was also trying to convey that I want to study performance optimization and low-level programming languages during my master’s degree, which ended up as “programming that’s closer to the hardware”…)

Recently, I read Sophia Amoruso’s autobiography, #GIRLBOSS, which details how she went from being a broke twenty-two-year-old with no college degree to a multimillionaire entrepreneur running her own clothing e-commerce website, Nasty Gal. The book itself is not outstanding—as you may guess from the title, it drips with a cishet white feminism that was all the rage in 2014, and the prose is unrefined—but reading it made me realize why Nasty Gal became such a success, rising above so many other fashion e-commerce websites of the late aughts. Amoruso worked extremely hard, yes, but the magical, unattainable factor that made Nasty Gal famous was Amoruso’s taste.


How does one develop good taste? The following rules seem to hold true:

  1. Having great taste in one domain does not indicate much about your taste in any other domain. You can have terrible taste in music but good taste in friends, or vice versa.
  2. As a corollary to #1, it’s fine to have bad taste in many domains, but it seems important to have good taste in the people you surround yourself with, and good taste within your career.
  3. Taste can be developed through experience and exposure.
  4. When you’re young (and I’m still young, too), you may not have had the experience and exposure to develop your own taste, so you may mimic the taste of those around you or those you look up to. This is fine to some extent, but I’ve seen many people “wake up” during or after college and realize that they never really wanted the life they’re now living.
  5. If you’re not certain if your taste is good or not, observe the reception to what you do. More on this later.

Does AI have good taste? Well, it is trained on the corpus of human society, and presumably since society has deemed certain artifacts “better” than others, AI can learn what is “good” and what is “bad”. But it relies on what it has seen before; I’ve never been pleasantly surprised by some clever phrase churned out by ChatGPT. AI can produce many options lightning-fast, but does it know which option is the best?

Recently, I received a discount code to try an AI portrait generator. I uploaded fifteen pictures of myself, and after an hour, the website spat out two hundred options. None of them were usable, and some looked pretty weird.

ai generated image

something doesn’t seem right…

Anyway, I don’t doubt that, in the near-future, advances in computer vision or machine learning would allow a program to sort through all these pictures and toss out the ones that are obviously anatomically incorrect, such as the one above. But figuring out what is a fantastic photo vs. a merely good one—that still seems out of reach.


As mentioned earlier, sometimes your taste is bad in a way that matters, and it’s better to realize that sooner rather than later. I’ve known startup founders that worked on the same idea for years, even though they were never able to garner an audience, and while there are many reasons why this could happen, from my understanding of the situation, a major factor was that they couldn’t figure out what others actually wanted. That’s what taste is—discerning what others actually want, whether that’s sorting through two hundred AI-generated photos to find the one that will look nicest on social media, or choosing a business idea and plan that will appeal to customers, or blending the right melody and lyrics and hook to produce a pop hit.

Seems hollow to base your taste in what others want, but does taste really matter if you aren’t making decisions that rely on others’ opinions? If you’re making a song for yourself, it’s perfectly fine if it’s bad. And yes, there are many visionaries who are “ahead of their time”—their resolve in pursuing their craft despite the lack of outside validation is incredibly admirable. If I were them, I would have no idea if my taste was bad or simply too revolutionary for the era. And in this late-stage-capitalism society, it feels pretty much impossible to cling to unrecognized genius unless you’re already independently wealthy.

Cross posted on Substack here