Just as every business plan needs to mention AI in the early days of 2024, so too do major new tech products. And nowhere is that more in evidence than in this year’s Consumer Electronics Show (CES). Manufacturers are putting AI in everything.

Even a pillow.

The Motion Pillow uses AI to detect that you’re snoring and adjust your head position to stop it. And it’s available to order right now, which is why I’m glad my wife doesn’t read NEXT Insights, as she’d be ordering one for me right away…

Occasionally, this pushes to a ridiculous extreme. You can have an AI help you cook a steak, for example. That feels like the AI version of the infamous internet-connected fridge. They still exist, but they’re still far from the mainstream of the product.

But this is machine learning-based AI of a type we’ve seen for years. What’s happening with generative AI and large language models (LLMs)?

AI conversations with your car

Well, you’ll find them in your (future) car. Amazon and BMW are working together to bring LLM-based AI to cars. Again, at first glance, this sounds like a terrible idea:

“I’m sorry, officer. The crash wasn’t my fault, my car was hallucinating…”

The reality is a little bit more prosaic — and a lot less worrying. Their solution aims to give you voice control of car settings and functions. Over the past decade, there’s been a battle between the desire to make everything screen-based — like Apple’s new iteration of CarPlay — and the functionality of a knob or dial. The latter have the advantage of being usable without you having to look at them: the physical affordances matter. Using Alexa with an interface specifically trained on the car’s features makes for genuine hands-free control of the car, keeping your hands on the wheel and your eyes on the road.

Volkswagen is doing something similar with its ChatGPT integration. You trigger the AI agent by saying “Hello Ida”, immediately giving everyone over a certain age nostalgic flashbacks to Knight Rider and its eponymous hero talking to his car KITT.

RabbitOS: the large action model?

One hardware product announcement — and a largely unexpected one — that has caught plenty of people’s attention is the Rabbit R1. It’s a little, handheld AI assistant, that aims to be more than a conversationalist. It’s using what the team behind it are calling a Large Action Model to go beyond conversations into delivering actions based on voice requests. And it does this because it’s been trained on app interfaces to understand how they work.

Here’s how the news release describes it:

LAM is designed to make AI systems see and act on apps in the same way humans do. It learns by demonstration – observing a human using an interface and reliably replicating the process, even if the interface is changed. It works by first understanding complex human intentions, then spinning up on human-oriented interfaces across all mobile and desktop environments on a customized cloud platform, and finally interacting with apps within the platform to achieve certain objectives without the need for complex, custom integrations like Application Programming Interfaces (APIs).

This does mean that you need to set it up and configure it via a web portal, and then give it access to the apps and services you use. It aims to become a universal interface in front of those apps, so your intent is the defining thing about the device, not app interfaces.

Oh, and it’s pretty cute:

Here’s the intro video, if you want a deeper idea of how it works:

It’s available for pre-order for $199, which is within many people’s “let’s have a play” threshold. It ships in late March in the US, and globally later in the year.

The AI-powered robotic ball

The Rabbit R1 appears, at least, to be real. But, as ever with CES, some of the announcements are very much vapourware. Back in 2000, Samsung announced Ballie, a mobile video streaming and smart speaker, that never shipped. Ballie is back for 2024. He’s gained a few pounds, but with the girth has come AI smarts:

Of course, it remains to be seen if this version of Ballie will make it to the market.

The year of applied AI

But what’s very clear is that everyone’s looking for how they can incorporate AI into what they’re doing. After years of false starts and technologies looking for a reason to exist, from blockchain to the metaverse, AI looks to define 2024 as companies partner with the LLM providers, and seek ways of delivering customer value by integrating it.

The way forward won’t be straightforward, and some early pioneers are stumbling. Humane’s AI Pin is available for pre-order, but the news of layoffs at the company doesn’t suggest that those pre-orders are going gangbusters. While the Rabbit Ri is $199, the pin starts at $699 with a monthly subscription on top…

Meanwhile, the big name in AI, ChatGPT, is facing a major lawsuit from the New York Times, one that it might struggle to see off. But, in the background, it’s also working with former über-Apple designer Jony Ive and his firm LoveFrom to design hardware implementations of their LLMs. And they’re poaching key Apple designers to do it.

If 2023 was the year of Generative AI expressed through software, 2024 is shaping up to be the year when it gets hardware all of its own.

Photo by Cwiela_CH on AdobeStock.