On Creative Judgment

Thoughts on AI and Turning Pro.

Image generated using OpenAI’s DALL·E 3.

One of the opportunities I’m most excited about with AI is its ability to amplify creative insight.

Years ago, I heard a passage from an interview with Ira Glass that really stuck with me:

All of us who do creative work, we get into it because we have good taste. But there is this gap. For the first couple years you make stuff, it’s just not that good. It’s trying to be good, it has potential, but it’s not. But your taste, the thing that got you into the game, is still killer. And your taste is why your work disappoints you.

A lot of people never get past this phase, they quit. Most people I know who do interesting, creative work went through years of this. We know our work doesn’t have this special thing that we want it to have. We all go through this. And if you are just starting out or you are still in this phase, you gotta know it’s normal and the most important thing you can do is do a lot of work.

In other words, taste is what gets you into the game, but it’s your ability to make that keeps you in the game. This holds true for many creative pursuits — from design, painting, and music, to writing, coding, and entrepreneurship. All beginners face a gap between idea and execution — and often give up before their work ever gets a chance to meet their own standards¹.

Now picture a young graphic designer who would be able to generate ten different landing page designs and use their creative flair to pick the best one. Or an architect who could automate all the tedious, detail-heavy components of a new housing complex to better focus on the bigger vision.

This is the promise of generative models. Through the familiar interface of language, they weave our creative insights into tangible outcomes — turning ideas into reality. Powerful tools like ChatGPT, Midjourney, and Gen-2 break down the barriers to creating advanced text, code, images, music, and video, making professional-grade artistic and technical production available at scale². Echoing Ira Glass, if the most important thing we can do to turn pro is “do a lot of work”, then generative models speed up the process by shortening the feedback loop and allowing us to fail faster — effectively removing the barriers between us and the work³.

This has several implications:

• The value of creative insight is going to increase. Concurrently, the value of technical skill (the manual act of writing, designing, coding) is going to decrease.

• The best curators are going to become the best creators. Curation, which involves the request, careful selection, and approval of outputs using one’s personal taste and discernment, is going to become the default way of creating. This will lead to shorter, more specialised trend cycles.

• Communication and empathy are going to become essential skills. The ability to translate creative insight into words is going to be key to harnessing the full capabilities of AI and producing high-quality outputs⁴.

Ultimately, the process of making is more art than science. In the age of AI, our best shot at producing work that meets our own standards, especially in areas where our taste is already killer, lies in inundating ourselves with examples and quickly being able to say: “try again”.

1. This provides a possible explanation of the Dunning-Kruger effect, which occurs when beginners sometimes overestimate their own competence in new areas. My working assumption is that we are prone to overestimating our abilities in areas where our judgement is already good, mistaking our intuition of what is good and bad for the actual skill of producing the good and avoiding the bad.

2. A second-order consequence of this is the (re)birth of making-as-a-service. In the same way that travel agents once connected travelers to service providers, professional makers currently act as intermediaries between the human experience and the creative realm. I argue that GenAI is going to change this by turning us all into makers.

3. I recognise that this focuses heavily on the outputs. I am not denying the importance of learning and practicing from first principles on the journey towards mastery. There is infinite value (and satisfaction) in the creative process itself — after all, the painter falls in love with the paintbrush long before they are able to produce a masterpiece.

4. I expect experts will still have the upper hand here, because they’ll know what to ask (the technical terminology and concepts that beginners lack). I see it with programming — ask ChatGPT to build a generic machine learning model, and its outputs are unusable; provide targeted instructions with the right level of detail and context, and it works wonders.