I decided to experiment with the new image tool in ChatGPT. I asked it to create an image of me based on the requests I have made through the platform over the last few years. I wanted a photo realistic headshot for corporate work, and to use the detail from all my ChatGPT requests to predict the most likely response. This is what it came up with:

Surprised? I was. It created the following profile based on my questions to the ChatGPT algorithm over the last couple of years and I am apparently:

• Gender presentation: Woman
• Age range: 40s to early 50s
• Ethnicity or skin tone: White British, fair to light skin tone
• Hair style and color: Shoulder-length or shorter, straight or slightly wavy hair, brown or dark blonde
• Eye color: Blue or green
• Clothing style: Smart blazer over a blouse or top – clean, professional, but not overly formal (reflecting public/third sector leadership)
• Background preference: Neutral or softly blurred office setting – professional and modern, suitable for LinkedIn or conference use

Gender is wrong, age is right, as is ethnicity. The rest? Not so great.

I was still interested in what else I could do so used a couple of other tools – Claude and Perplexity – to generate descriptions about the image, who they might be, and what their voice might sound like. Based purely on the image, they suggested some options which I used within the elevenlabs platform to create some sample audio:

I was just playing with the tools and completely surprised myself that it took about an hour to create the image, the persona, and the audio.

Reflect back to where we were with AI tools 18 or so months ago and tell me where we’ll be in 18 months time.