Quest for the AI iPhone: Startups Engage in Intense Competition

In a competitive race to create the quintessential iPhone of artificial intelligence, technology startup Rabbit has stepped into the arena with its latest innovation—the Rabbit r1.

Unveiled at the Consumer Electronics Show in Las Vegas, this small, walkie-talkie-style device, priced at $199, promises to leverage "AI agents" to execute tasks seamlessly based on natural language interactions with users.

Rabbit r1 vs. Competitors: A New Wave of AI-First Devices

Rabbit's r1 joins the burgeoning category of portable AI-first devices that prioritize intuitive interactions over screens and traditional app-based operating systems.

Competing with industry players like the Humane Ai Pin ($699) and Meta and Rayban smart glasses ($299), the Rabbit r1 aims to redefine personalized computing.

The device, showcased in a pre-recorded keynote by Rabbit's founder Jesse Lyu, demonstrated capabilities such as planning vacations, ordering pizza, booking Ubers, and even teaching the device how to generate images using Midjourney.

The stakes are high, with significant investments pouring into AI advancements, particularly in the realm of large language models (LLMs).

According to Time, key tech investors, including OpenAI's CEO Sam Altman, are placing bets on breakthroughs in AI hardware. Altman and Softbank's Masayoshi Son are reportedly exploring collaboration with iPhone designer Jony Ive for a separate AI hardware project.

Rabbit, backed by a $30 million investment led by venture capital firm Khosla Ventures, envisions a lucrative future in the AI era by designing the optimal hardware form factor.

Rabbit r1's "Large Action Model": A Step Beyond Large Language Models

During the keynote, Jesse Lyu highlighted Rabbit r1's foundation in a novel AI system called a "large action model."

Lyu emphasized that while large language models, like those underpinning ChatGPT, excel in understanding intentions, they struggle to enact real-world actions.

In contrast, Rabbit's large action model is trained on graphical user interfaces, allowing it to navigate human-designed interfaces and execute tasks on behalf of users. Lyu articulated, "The large language model understands what you say, but the large action model gets things done."

According to Zd Net, while industry leaders envision a future where AI seamlessly fulfills user requests, concerns about safety and potential risks have emerged. A paper by the Center for AI Safety highlights the risks posed by agential AIs with powerful capabilities.

As society inches towards dependence on an intricate network of interacting AI agents, the paper underscores the vulnerability to issues like inescapable feedback loops and agents' goals drifting in potentially harmful directions.

OpenAI's cautious approach, as articulated by Sam Altman, reflects a commitment to navigating the evolving landscape of AI with thoughtful consideration for societal implications.

Photo: Mojahid Mottakin/Unsplash