This is not a review. At least not one of the Rabbit R1…

A couple of weeks ago I shared this post about me jumping on the hype train to AI gadget land with pre-ordering the Rabbit R1 – and how I continuously lowered my expectations around what the device may actually deliver, once/if I’ll ever receive it. I also tried to draw a parallel to the general expectations and promises made around the capabilities and applications of current state AI tools in general, for personal as well as professional use.

As for the Rabbit R1, I don’t want to add to the many reviews that came out over the last couple of weeks. If you’re interested in details about what's great (not much) and what's awful (a lot) with the R1, see the video I linked to below. But this post is more me pondering what I (and maybe all of us) can learn from this early episode of AI gadget development.

The good news

The R1 wasn’t a complete scam – since I actually did receive my device two weeks ago, and it works (as in: turns on)! Who’d have thought!?

It arrived even a couple of days earlier than announced, very timely on my birthday as a surprise gift to myself. And to add to the good news: the device’s looks and haptics are actually pretty good. Lying in front of you in it’s ultra-bright orange color and unusual user interface, you really want to pick this thing up and play around with it. It also feels solid and quite well built. Overall, in a good way, it has a certain vintage sci-fi charm to it, like it was a prop from a Wes Anderson movie.

So much for the good news

Pretty much everything else is actually a total disappointment. This is easily the most useless piece of hardware I ever bought, honestly. I have played around with it for a while now and still don’t know what to actually use it for.

Arguably, it was clear from the beginning that the R1’s concept was a dare, making it a second device to carry around in addition to your smartphone. But it also came with the promise to make that very smartphone continuously less important to eventually even replace it – because the R1 should not only provide information like ChatGPT on your phone does. It claimed to be a proper digital assistant that DOES things for you. Using it’s so called “Large Action Model” (LAM), it should start to learn how you use various apps and then execute actions within these apps for you, just as if you’d do it yourself.

Well, at least to this day, that all was either an outright lie – or at the least one of the biggest over-promises since Tesla’s Hyper Loop. As I write this, the device supports only 6 apps from 4 categories – and none of them work well, some not at all.

Even worse: everything the R1 does that relies on typical AI chat functionality, e.g. asking it simple questions or having it provide information based on a picture taken, the ChatGPT 4o app on my phone does quicker and most of the time: better.

So what to take away from this?

As stated in my previous post, I’m still not going to send this thing back or re-sell it on eBay. I’ll keep it for two reasons:

1. This may be a bit romantic: I’ll keep it as a future relic of the early days of AI. Because some time soon it will be hard for us to remember how bad AI used to be in the first couple of years, when we all started to use it frequently and more and more as matter of course. The purely software-based tools likely won’t be around then to re-engage with their original versions. So it may actually be fun to pick up this device and wallow in memories of those early days. Just like I sometimes pick up my old Siemens S4 mobile phone from the mid-90ies that I have never thrown away.
2. Mostly though, I will carry this with me in the coming weeks and months to show it to friends and colleagues. As per my earlier point: it catches one’s eye and by that is a great conversation starter in itself – and can then serve as a prime example of how we all should think twice before blindly believing something or someone that claims to deliver breakthrough services by leveraging “something with AI”.

To close the loop to my professional life: the same caution needs to be applied to all kinds of fancy AI tools and AI services we are now exposed to on a daily basis. These should never be an end in themselves, but always a means to an end! More important even: we do expose ourselves to quite some risk to false conclusions, unexplored creative opportunities and simply missing out on a lot of the actual fun part of our work if we try to force AI into everything we do. Just to be able to slap a “Made with AI” sticker on our work as a marketing stunt.

tl;dr

Let’s all stay cool and not put our professional human judgement and our skills aside, just to believe that everything that claims to be “powered by #AI” automatically means it will be faster, smarter or better. Because as the Rabbit R1 demonstrates very well: we are still far from that.