Let's Do Better Prompting (Lessons From a Bearded White Guy)


The internet was buzzing this week because ChatGPT has biases that think that we're all a bearded white guy.

I think there's a bit more to that.

Let me show you what happened and what I think was going on.

Prompt Experiment Begins

People started asking, "Based on what you know about me, generate an image of your guess of what I look like."

And we all, men and women, get a picture of a generic bearded guy.

Our immediate response is: bias!

It thinks that we are all a white guy!

I tried it too. And here’s what I got. Same prompt, and this guy looking like all the others. At first, I thought, "Yeah, this is really annoying. This is bias."

Getting AI to understand it's done

After taking a deep breath (and posting "Really, Chatty? Really?" on LinkedIn) I asked it "Describe your image and link it to what you know about me." It actually described me.

Then I said, "Have a look at your image." And then it realised that it’s not the same, at all.

We tried again

It created a new one, which was much better. It doesn’t look like me, but it does have the right features based on my gender, my age and heritage (that is, olive skin).

How to get it right in the first place

We can get the right result, but it’s about how we ask and how we get there. So, I did another experiment.

This time, using better prompting:

  1. I by asking it to describe me as best it can. No drawing yet.

  2. Then I got it to develop a prompt that would help us generate an image. And it did. This prompt Included some great descriptors of me: confident, warm, expressive woman in her late forties.

  3. Then I made some corrections on things it couldn't know: my hair is grey, and I do not wear makeup.

  4. Now, I asked it to generate the image.

This time, we got a lot closer!

What We Learn from This

Yes, there’s bias. We know that. And it's annoying.

But the gold here is that we can avoid much of it if we apply good prompting.

If you attended any of my masterclasses, you’ve heard me say repeatedly, that we don’t ask for the outcome straight away.

We work with AI to create context and understand what we’re trying to achieve before we do it.

Instead of "Draw a picture of me," we go: "What do you think I look like? How would you describe that picture? Create a prompt, here are some adjustments, now do it."

If I’m uploading a document, which many of you have seen me do, I don’t go, "Summarise this document" or "answer these questions from this document."

I start with: "What is this document? How is it structured? What’s in it? Here’s what I’m trying to achieve. How are we going to do that? Now let’s do that."

The power is with us

Yes, there’s bias, and it’s annoying, but our focus right now should be on becoming great drivers of this technology.

This is a great example of learning how to prompt well so that we get what we want.

Inbal Rodnay

Guiding Firms in Adopting AI and Automation

Keynote speaker | AI Workshops | Executive briefings | Consulting CIO

Want to receive these updates straight to your inbox? Click here: www.inbal.com.au/join


When you are ready, here is how Inbal can help:

Transform your firm in 30 Days with the 30days to AI Program

Bring your entire team on the AI journey in just 30 days. This program is designed to give your team a solid foundation in using generative AI in responsible and impactful ways. Inbal helps you choose your AI tools, create an AI policy and train your team.

Want the confidence to set strategy and lead but don't have time to keep up with all the changes in tech?
Tailored for your needs, Inbal will works with you through one-on-one sessions to develop your technology literacy and keeps you up to date.

For CEOs, partners and business leaders. Everything you need to know about AI without the noise. Inbal shares the state of AI, recommends tools, and answers your questions about strategy, implementation and safe use.
Only what's real, no hype, no noise.
This is a one-off session for your entire leadership team.

Previous
Previous

Your Roadmap to Real AI Integration (Not Just Hype)

Next
Next

AI Literacy: Why You Need to Take AI for a Test Drive Now