AI assistants and the battle for user trust

Thirty years ago, children played with a digital pet named Tamagotchi, which communicated its needs and required humans to adjust their lives around it. Recently, Meta unveiled Muse Charm, a small keychain device that houses Meta's personal AI agent. This marks a significant step towards achieving the long-sought goal of a true personal assistant in computing.

Unlike previous chatbots, copilots, and voice assistants, this AI personal assistant will be more like a chief of staff or an alter ego: persistent, personalized, aware of your past and priorities, capable of acting across your work and life, and increasingly skilled at anticipating your needs even before you request them.

Before Muse, there was Instinct, an invite-only assistant launched by 23-year-old Noah Shinn, which gained a cult following this summer. It integrates with various personal systems, such as email, messaging, calendars, and more, and can be contacted via text, WhatsApp, or phone. Users can ask Instinct to perform tasks like booking hotels, organizing inboxes, arranging rides, canceling subscriptions, or following up on life's loose ends, without needing to navigate through settings or apps.

This development signifies a shift in computing from "finding the right app and operating it" to "telling someone what you want done." Meta's Muse extends this concept to the mainstream market, offering connections to email, calendars, services, web browsing, appointment booking, form filling, shopping, long-term goal tracking, and post-app closure continuation.

Microsoft's Clippy, Google's Siri, Google Assistant, and Amazon's Alexa were early examples of AI assistants, with varying levels of usefulness for tasks like setting timers, weather updates, and music playback. However, generative AI has brought language and reasoning capabilities to these assistants, turning them into agents with hands, enabling them to perform actions beyond simple tasks.

As personal assistants gain more power and access to sensitive information, the question of trust becomes paramount. While Meta claims that Muse uses secure credential storage, virtual machines, and permission controls, users may still be hesitant to grant such extensive access to their devices due to Meta's history of privacy controversies. Additionally, platform conflicts may arise as different AI agents vie for control over various aspects of users' digital lives.

Ultimately, the focus may shift from the best chatbot to the most trusted personal intelligence, residing between users and the digital world. This transformation in the role of computers suggests that the future of personal computing may not revolve around personal ownership of devices, but rather around having an AI identity that understands and assists us everywhere.