Apple's New Siri: Early Tests Show Mixed Results

Early evaluations of the forthcoming Siri AI for macOS reveal a system with significant potential, though it currently exhibits notable limitations. Introduced in a developer beta, the AI is still undergoing development, with ample opportunity for refinement before its wider release later this year.

While Siri AI excels at tasks involving Apple's native applications like Photos and Messages, it struggles with data from non-Apple services such as Signal or Google Photos. The AI's ability to analyze data, such as calculating average benchmark scores from screenshots, is promising but inconsistent. It can differentiate single-core, multicore, and GPU scores but is often disoriented by mixed test types or extraneous data.

Users have noted that Siri AI's visual intelligence, capable of interpreting on-screen data, is restricted. It cannot perceive all data if it exceeds the visible screen area in applications like Google Sheets. Attempts to use Siri AI for automating complex tasks, like running benchmarks, were unsuccessful as it cannot perform actions within applications. This indicates that while the AI can launch apps, its capacity for intricate in-app operations is yet to be realized.