Rabbit's r1, the AI assistant gadget whose hype train has slowed since its introduction at CES, has some changes to announce — but they are unlikely to sway its detractors' minds just yet.
A new "beta rabbit" mode enhances the device's conversational AI capabilities, particularly in difficult or multi-step directions. It should also be better at asking clarifying questions when it is uncertain about something.
Other suggested possibilities include creating a travel schedule and looking for offers or product recommendations. As occasional chatbot users ourselves, we've discovered that this type of activity is impressive in a demo but rarely effective in practice.
For example, chatbot itineraries can be strange and unpredictable, and comparing specs and costs demonstrates web-scraping ability but is ultimately cumbersome on such a little screen. Who is going to believe such poorly sourced book recommendations?
The new alarm and timer features, which may be found here, are welcome, but can sometimes cause delays. For example: "Set a timer for baking chocolate chip cookies": At which temperature? How many? Which kind of cookies? This is the recipe for culinary disaster. The AI cannot possibly be aware of such information. On the other hand, it is totally fair to inquire, "How long should two dozen chocolate chip cookies be in the oven at 300 degrees?"
Of course, everyone is waiting for the much-vaunted but so far exceedingly elusive "large action model" that the business promised back in January. The pitch, which I saw as aspirational but not ludicrous at the time, was that the model had been trained on phone and web app interfaces and could traverse them autonomously to complete user-selected activities. So far, such a feature has not been demonstrated outside of demos, and if an action is claimed to use it, it is indistinguishable from what an API or standard action scripting could accomplish.
I'm still enthusiastic about the long-term utility of this strange little gadget, which is why, despite the fact that I've had almost no reason to use it since we received one to review, I haven't tucked it away just yet.
I've asked Rabbit when we may expect news on the LAM and will update this page if I hear back.