Another week, another experimenting post.
Recently my mom read an article in the Washington Post about AI. She followed this by texting me that she was tired of AI. Dear mother, first of all, I hear you. Second, AI is excellent for many reasons, but one particular reason is its dependency on A/B testing. Therefore, at Experimenting, the newsletter, we love AI.
This week's post is a super quick light read. I hope you enjoy it. Thank you for reading.
I am an adult who eats too many croissants. Maybe you are too?
Sometimes a croissant is flaky yet soft, with infinite layers of joy. Other times, a croissant is dry, stale, and joyless. A bad croissant feels like sacrilege. A good croissant feels like a reprieve from reality.
I am motivated to find the best croissants this city has to offer. This will take research, many many weeks of research. To accompany this effort, I’m also writing a detailed report evaluating the croissant variants.
In a few weeks, you’ll see this evaluation in your inbox, but until then, I decided to finally authenticate my Google account with OpenAI. Let’s see what ChatGPT thinks of this experiment of mine, shall we?
OpenAI: ChatGPT
I asked ChatGPT — “can you a/b test croissants?”
See the following image for the response.
Thanks, ChatGPT. Of course we can A/B test croissants, just as any other product. The language model continued to churn out the general steps in the A/B testing process - define your objective, define metrics, split your audience, etc etc.
ChatGPT’s response was underwhelming and yet serves as an example of why this machine learning model can't replace good writing by real humans. Imagine how entertaining it would be to read a detailed article depicting the steps to evaluate croissants in the scope of an A/B test.
Let’s move on to DALL-E.
OpenAI: DALL-E
I asked DALL-E to generate an “oil painting of a cat A/B testing croissants from a park bench.”
See the images that follow.
I actually have a liking for these illustrations.
If only there were multiple versions of croissants to depict an actual A/B test. How to improve these images…well maybe DALL-E couldn’t interrupt the intent of ‘ab testing?’
Let’s modify the input slightly.
The following images were generated from the text “a futuristic cat experimenting with croissants on a park bench.”
Croissant variants, cats and OpenAI aside, the beauty of A/B testing is that it can be applied to all aspects of life, not just software and product development. Because really, life, in some sense, is a series of many experiments.
I hope you have a wonderful week.
Thank you for reading.