Ask Better Questions


AI is having a moment.

The past year has seen an explosion of awareness and interest, mainly due to ChatGPT, but also because of other generative AI tools like Dall-E, StableDiffusion and Lensa. Each new tool has been celebrated in media, along with some hand wringing over how the tools may displace creative people who produce text or images for a living.

In the heat of worrying about how AI will change the job landscape, or whether automation will make anyone redundant, I suggest that we are thinking about this wrong:

  • Will I lose my job?
  • Will AI make my skills redundant?
  • How can I possibly compete?

Instead, consider this POV for the current moment:

  • AI is suddenly relevant, not because it is solving problems, but because it seems to be solving problems by automation.

Some Humans Required

AI-enabled processes require humans at every turn.

We must recognize the current AI offerings for what they are - a set of tools that can be used to improve the speed of delivery. These tools don’t work without humans yet.

Everything produced by these generative tools was driven by human input.

Some person asked for that photorealistic image of a teddy bear on a skateboard in Times Square. The transaction requires two things in addition to the generative AI tool to produce the image:

  1. Desire for an image
  2. Creative input from a human

Without these inputs, the tool does nothing, just as a ball of yarn cannot knit itself - it needs a pattern, and a person to become a sweater.

Seen this way, generative AI is a set of power tools that can produce magical results, but are just tools that automate activities humans have been doing slowly.

Creativity Cannot Be Automated

Prompt:

What does creativity entail?

ChatGPT:

Creativity is the ability to produce new and original ideas, concepts, or things. It involves using imagination and original thinking to generate new ideas, and often involves breaking away from conventional thinking or traditional ideas. Creativity can be applied in various fields such as art, music, writing, design, science, and more. It can also refer to the ability to solve problems in an innovative or unique way.`

I think that’s a pretty good dictionary definition, but it’s dry and lacks anything that would excite the reader.

Prompt:

Give a metaphorical example of creativity in action.

ChatGPT:

One metaphor for creativity in action is a garden. Just as a gardener carefully cultivates a garden, nurturing the seeds and tending to the plants as they grow, a person who is being creative is also carefully nurturing their ideas and allowing them to grow and develop. Just as a gardener must be patient, persistent, and willing to take risks, a creative person must also be willing to take risks and think outside the box. And just as a garden can yield a beautiful and diverse array of flowers and plants, creativity can yield a wide variety of new and original ideas, products, and solutions.`

Hmmmm.

Prompt:

Make a short example using this garden metaphor.

ChatGPT:

A young artist, much like a gardener, starts with a blank canvas, and begins to plant the seed of an idea, carefully nurturing it with their imagination and skill. As the work progresses, the artist must be patient and persistent, tending to the piece as it grows and evolves, just like a gardener tends to their plants. And just like a gardener might experiment with different seeds, colors, and textures to create a beautiful and diverse garden, the artist plays with different techniques, mediums, and styles to bring their unique vision to life. The end result is a stunning, one-of-a-kind masterpiece, much like a beautiful garden full of vibrant colors and rich textures.`

This has all the creative charm of a copied Wikipedia article. And that’s actually the point, because generative AI will generate combinations of text or images that it has seen before, and while the result may be surprising, ultimately it’s still an automated approach to choosing among a lot of possibilities.

Choosing among possibilities is how humans find answers too - in fact, the series of prompts above is a typical experience of a person using generative AI to produce a result. The iterations are where the creativity lies, as the human revises the question across several prompt requests, seeking a generated result that matches their goal.

Automation makes it faster to do this, and digital automation in particular makes it faster because there seem to be no wasted materials - no stack of bad painted canvases, or wastebaskets full of crumpled paper to document the artist’s path.

Creativity Benefits From Constraints

Prompt:

Write a haiku using this garden creativity metaphor.

ChatGPT:

Seed of thought planted / Nurtured by imagination / A garden blooms forth.

This is better, because it feels creative. It’s concise where the previous results were prolixious.

Prompt:

Write something creative based on this conversation. Be concise.

ChatGPT:

The artist’s canvas, a blank slate, A garden waiting to be created, With each stroke of the brush, A petal, a leaf, a stem take shape, A world of color and form blooms forth.

This is worse. It lacks the spark that we recognize as creativity, because it is just a machine churning out combinations of predigested text. The magic disappears a bit once you see what the tool is really good at - collating and summarizing examples it has already seen.

What we might imagine to be creative output is exposed as an illusion. It’s not nonsense, but it isn’t what we wanted.

Just as an IMAX movie screen can trick your senses into feeling present in the image, a generative AI tool that has the entire text of the internet can trick you into thinking it is creative when it’s merely fast at creating similar examples.

Creativity Requires Selection

The last two examples demonstrate how attempting to constrain the output can produce entirely different results, some of which may be much worse. Choosing among a set of outputs is how we move forward.

Creativity is a selection process as well as a combination process. We choose among a set of ideas, materials, and techniques, and then combine them. If we don’t like the result, we try again, with a change in idea, material, or technique and see the next result.

Iteration with intent is the essence of creative activity.

Shoulders of the Crowd

Generative AI uses the predigested text and images of much of the Internet. In essence, it is the internet, but in chat format.

Our egos lead us to think in terms of how we are separate, different, better/worse than our fellow humans. Tools like GPT will allow most humans to produce ‘better’ text than they could previously, but the challenge of actually creating new ideas and connections remains.

Some better questions

  • What do people actually want?
  • What do I enjoy creating?
  • What can I create that cannot be automated away?

Beyond these generic questions, ask yourself:

  • What is my unique value proposition?

If humans are required, then the trick is to ensure that your talent can be enhanced by automation, or that your audience isn’t interested in things that can be automated.

Self awareness and intent are the personal tools we must bring to this new game of automated creation.


See also