Google’s new hidden search feature

You can generate images within Google search, but the feature is picky and hidden.

With Google's SGE turned on through Google Labs, you can prompt Google to: Draw a bike with bells and whistles.

This feature dropped last Thursday but there isn't much information about it online.

I’m still experimenting, but it’s particular about how you frame your prompt.

You can generate images by beginning your prompt with:

  • Draw

  • Sketch

You cannot generate images by beginning your prompt with:

  • Generate

  • Ideate

  • Create

  • Model

  • Illustrate

  • Spawn

  • Make

  • Design

 

You cannot generate images without Draw or Sketch in your search prompt.

For example: A bike with bells and whistles. will not generate images. You must begin the prompt with Draw or Sketch.

 

I haven't discovered other words that work.

Google's image generation, powered by Imagen, is nowhere near as good as Dall-e, Midjourney or Adobe Firefly, but this sets an interesting precedent: Google wants to be the only place you go to for images.

If you search for an existing image online, you already go to Google. But to generate an image, you have to go somewhere else.

This may be an imperative beginning of Google bridging the gap between what people are already doing vs what people need to learn about generative AI tools.

Google is tackling the problem of minimizing the change in user behavior by allowing users do what they've always done to find a picture Google it.

The only behavior difference now is remember to add Draw or Sketch to the beginning of your search.

The majority of people don't need all the great design capabilities that competitive image generators include in their user interfaces. People want something familiar and intuitive and this is the best solution to that problem so far. The images it generates just need to get better.

Previous
Previous

Putting a price on value

Next
Next

Feature creatures