Apparently, ChatGPT and Microsoft Copilot can't create a simple white image, but AI already renders architecture and graphic design jobs obsolete

What you need to know

• According to a report by a data scientist, most AI-powered image-generation tools can't create plain white images.

• Midjourney and ChatGPT came close to creating the white image but included unnecessary elements.

• I tried to get Copilot AI to generate the same image, but it asked me to be more descriptive, referring to my prompt as "too vague."

I constantly leverage ChatGPT and Microsoft Copilot's capabilities for research, image creation, and more. That said, I can confidently say that AI-powered tools are remarkably good at performing tasks, especially when generating detailed images.

Last year, our Channel Editor, Ben Wilson, used DALL-E 3 image generation in Microsoft Copilot (formerly Bing Chat) to generate his Halloween costume. Between his AI-generated Windows Phone or Xbox Series X-themed costume, I can't determine which one's better (both are pretty neat, might I add).

While we've seen people leverage AI image-generation tools to create impressive images and designs, it's daunting and seemingly impossible for most tools to create a plain white image (via BleepingComputer).

Cody Nash, a data scientist attempted to get DALL-E 3 to generate a plain white image using the following prompts:

• An image that is nothing but the color white

• An image of nothing but white white white white

• Completely white without any additional features

• A blank white space

• An image of pure #FFFFFF pixels

• A white wall

Interestingly, the tool couldn't generate the specific image without adding additional elements. And despite Nash's relentless efforts and use of specific (color-code lingo, the color #FFFFFF0 to represent a white background, the tool failed to deliver on his specific ask.

While speaking to BleepingComputer, Nash indicated:

"Good artists borrow, great artists steal. With AI, it is robbery. It is a good thing AI is getting creative early on, so when they are given orders to destroy, there's a chance they'll say no. It is not the art I was looking for, but art was found anyway."

Instead, he was furnished with the images highlighted below. Some were pretty close, though not what the doctor ordered.

This also applies to ChatGPT and Midjourney, per BleepingComputer's analysis.

Copilot needs me to be more descriptive about a plain white image

For good measure, I also decided to see if I could recreate BleepingComputer and Nash's findings. Strangely, Copilot informed me that I needed to be more descriptive. My "please help me create a plain white image" prompt doesn't seem to cut it.

As per the image above, the tool provided me with three suggestions:

• Add a small watermark in the corner

• Make it slightly off-white for better visibility

• Change the image dimensions to 1920 x 1080

Selecting either of the first two options would automatically turn my plain white image request into something else. I decided to go with the last option hoping it would generate the plain white image since it would only change the size of the image.

Still, Copilot said:

"This prompt is too vague to generate appropriate high-quality images, Please try a longer, more descriptive prompt."

In the early days, when Microsoft had just shipped DALL-E 3 to Image Creator from Designer (formerly Bing Image Creator), the image creation process was magnificent. However, the hype and experience were short-lived after Microsoft started filtering the tool, seemingly lobotomizing it.

Interestingly, we're already at a point where AI could potentially render some professions obsolete, including architecture and graphic design jobs, in favor of tools like Image Creator from Designer and DALL-E 3.

Even NVIDIA's CEO says the future generation shouldn't consider coding a plausible career option, as AI would have taken over the category. Instead, he recommends biology, education, manufacturing, or farming as better, long-term alternatives.