This past month has seen Meta AI expand its functionality across their products, integrated further within Instagram, with Story AI. It is a thing that is available outright to all Instagram users across most markets, making it one of the most singularly available AI tech, after WhatsApp. But availability does not equal quality.

This week saw the launch of Nano Banana Pro, Google’s newest AI model that allows image creation that has far surpassed anything we’ve seen in AI tech so far. It’s astounding to see how far generative AI has come in just a few years, to the point where it’s promoted AI video and literacy commentator Jeremy Carrasco (@showtoolsai on TikTok) to post a now viral video stating, “you will be fooled by an AI photo this week.”

Image courtesy @immasiddx on X.com

In his TikTok he talks about how Nano Banana Pro has evolved so much that generative AI media is so lifelike and realistic, that it will likely fool just about anyone. This AI intelligence also extends to their large language model as well, now becoming much savvier and has become harder to distinguish from non-AI based content.

Meanwhile, Meta AI has still a long climb ahead of itself. Anecdotally, I can talk of instances where I’ve been at dinners with friends these past weeks, and as a neat party trick I pull out my phone, and ask them to take a well-lit picture of me on Instagram Stories, and then continued to show them the new built-in AI features under ‘Restyle’, which allows you add, remove, or change features of the picture you’ve taken. It also allows you to select from a range of templates curated by Instagram. The AI integration does not extend to videos.

I selected from a (short) range of templated prompts to turn my outfit into a disco ball themed mirrored outfit. And while it did do it, it also made me appear a whole lot heavier set than I am. It generates two options from go, but both of those options didn’t appear to be any good.

I then selected the ‘change’ tool to restyle what I was wearing to be a pink Barbie hoodie, which it did generate convincingly so, yet with the typical AI fuzziness that came with the first wave of generative AI images. After further inspection of all the generated images, it truly didn’t generate anything that I – as a professional in the content space – would ever post to my discerning audience.

Which raises the question – who is Meta AI for? What purpose is it serving to the masses who now have AI capabilities?

Meta AI has become in a case study of becoming a novelty, good to show for friends and family who won’t taken it seriously, or might even have a laugh at its expense, or for the less savvy, lesser discerning user who is happy with the first-wave like AI results which appear to be as such.

Competing with Google’s latest AI iteration, and even Midjourney’s savvy image generation, Meta AI has a long journey ahead of itself to prove its value for the user, in an increasingly competitive market. It is an uphill battle where it is still swinging, but is constantly being beat by other, more useful AI tools.