Google Bard’s chatbot is now out of its beta period – so what do we know about Google Bard image input?
When OpenAI unveiled GPT-4’s image input it was a huge step. Examples of image input included turning a plan for a website drawn on a napkin into a functioning site.
Of course, we were excited to learn if Google Bard could do the same thing if given an image, and if Google Bard image generation was possible.
Since coming out of the Beta period, Google Bard has been updated with their massive PaLM 2 language model, dramatically increasing the parameters of the AI. You’ll now be able to load images into Bard thanks to Google Lens. Bard will now be able to analyze and describe the photo, and prompt Bard to write a caption for it, surely earning its spot amongst the best AI image generators.
Google has also announced that they have been collaborating with Adobe, meaning Google Bard will soon be able to generate images from text responses. There will then be the option to further edit those images in Adobe Express.
Originality AI detector
Can Google Bard understand images?
Before we jumped into sending Google Bard pictures of our favourite things – we first enquired if it could even understand images. Here’s what Google Bard responded with:
Effectively, Google Bard does understand images. However, it makes it clear that it does not understand images in the same way that we do. It’s important to remember this for later.
It can, however, learn from images and attempt to interpret them. This is an iterative procedure, so presumably, Bard’s image interpretation ability is only going to get better – which is definitely a good thing.
Can Google Bard create art?
Rather excitingly, Google is integrating Adobe Firefly into Bard. This means that when you ask Bard to create images from text, it will deliver the results via Adobe Firefly.
This will mean users can use Bard for a range of uses including social media posts, invitations, logos and more. Indeed, it looks like its features might easily rival AI image generators like Midjourney, or OpenAI’s DALL-E 2 which are associated with copyright infringement issues.
How do you send Google Bard images?
Like ChatGPT, Bard is a natural language AI, so we knew it wouldn’t be as simple as pasting an image into the chat. Here’s the result.
Bard states that there are three main ways to send it images:
- Upload a link to a file sharing service hosting the image
- Send the image as an email attachment
- Send the image as a text message
All seems fine at first, as sending a link to an image seems trivial for Google Bard. However, when we began to think about the second two options we were slightly perplexed. The generative AI states that you can send an email or a text message to its personal contact details, which is a little personal, right?
Naturally, we asked Google Bard what its phone number and email address are, and this is what it responded with:
Clearly, Bard gets a little confused at times, though it’s slightly amusing that it has seemed to forget that it is artificial intelligence at times. Or is it concerning?
Frequently Asked Questions
Does Google Bard understand images?
Google Bard doesn’t understand images as a human would, it told us. However, it can interpret and learn from images. Similar to GPT-4, artificial intelligence is now at a point where it can digest visual information.
Can Google Bard generate images?
Yes, with the latest update, Google Bard will gain the ability to generate images from text.