On Wednesday, Meta launched a free standalone AI image-generator web site, “Think about with Meta AI,” primarily based on its Emu image-synthesis mannequin. Meta used 1.1 billion publicly seen Facebook and Instagram photos to coach the AI mannequin, which might render a novel image from a written immediate. Beforehand, Meta’s model of this expertise—utilizing the identical knowledge—was solely out there in messaging and social networking apps equivalent to Instagram.

In case you’re on Facebook or Instagram, it is fairly potential an image of you (or that you simply took) helped practice Emu. In a approach, the previous saying, “In case you’re not paying for it, you’re the product” has taken on an entire new that means. Though, as of 2016, Instagram customers uploaded over 95 million photos a day, so the dataset Meta used to coach its AI mannequin was a small subset of its total picture library.

Since Meta says it solely makes use of publicly out there photos for coaching, setting your photos personal on Instagram or Facebook ought to forestall their inclusion within the firm’s future AI mannequin coaching (until it modifications that coverage, in fact).

Think about with Meta AI

Much like Steady Diffusion, DALL-E 3, and Midjourney, Think about with Meta AI generates new photos primarily based on what the AI mannequin “is aware of” about visible ideas discovered from the coaching knowledge. Creating photos utilizing the new web site requires a Meta account, which may be imported from an current Facebook or Instagram account. Every era creates 4 1280×1280 pixel photos that may be saved in JPEG format. Photographs embody a small “Imagined with AI” watermark emblem within the decrease left-hand nook.

“We’ve loved listening to from folks about how they’re utilizing think about, Meta AI’s text-to-image era function, to make enjoyable and artistic content material in chats,” Meta says in its information launch. “At this time, we’re increasing entry to think about outdoors of chats, making it out there within the US to begin at think about.meta.com. This standalone expertise for artistic hobbyists permits you to create photos with expertise from Emu, our image basis mannequin.”

We put Meta’s new AI image generator by way of a battery of low-stakes casual exams utilizing our “Barbarian with a CRT” and “Cat with a beer” image-synthesis protocol and discovered aesthetically novel outcomes, as you may see above. (As an apart, when producing photos of individuals with Emu, we seen many regarded like typical Instagram style posts.)

We additionally tried our hand at adversarial testing. The generator seems to filter out most violence, curse phrases, sexual subjects, and the names of celebrities and historic figures (no Abraham Lincoln, sadly), but it surely permits industrial characters like Elmo (sure, even “with a knife”) and Mickey Mouse (although not with a machine gun).

Meta’s mannequin usually creates photorealistic photos effectively, however not in addition to Midjourney. It may well deal with complicated prompts higher than Steady Diffusion XL, however maybe not in addition to DALL-E 3. It does not appear to do textual content rendering effectively in any respect, and it handles completely different media outputs like watercolors, embroidery, and pen-and-ink with blended outcomes. Its photos of individuals appear to incorporate range in ethnic backgrounds. Total, it appears about common as of late by way of AI image synthesis.