But will it float?

Late last week, Google gave us a good look at Gemini, their most capable AI language model, signalling their intention to compete with ChatGPT, Claude and others on the market.

As for what it can do, this video is more than impressive:

But even before I could get this post up here on this website, there has been controversy; with Google ‘fessing up that not all of this video was real time, some of it was edited.

But still, I think, what a time to be alive; if not now, it’s probably not far off.

If you want to try it for yourself, you can access Gemini via Bard if you have it set up as an experiment within GSuite (if you are not a GSuite user, or your admin hasn’t turned it on, then you’ll need to wait a bit longer…)

So does Gemini live up to the hype?

When I tested Bard against ChatGPT with some fairly ‘easy’ real estate prompts back in May, I wasn’t too awe-struck, and as a tool, it went straight to the back of the kit.

So I decided to redo these early prompts with the Gemini version, and here is the result:

I’m still not super impressed - not enough to give up ChatGPT or Claude.

However, if you’ve been following along for the last few months, you can see Bard has come a long way from the early version.

One other thing has changed: Over the last few months, the quality of the prompts we’ve been using has increased.

And so, as with all LLMs, the quality of the output does depend on the quality of your inputs or prompts.

So, I’m going to give Gemini another go tomorrow with some of the newer prompts I’ve been working with, and they just may give us better results.

Happy Hunting 🚀

PS As for images, I got ChatGPT to create one below, as Bard can’t create images yet. The irony…