Here’s an AI gadget that actually works — and one that needs improvement

The AI hardware shift: Major tech companies are moving artificial intelligence processing from cloud servers directly onto consumer devices, promising faster performance and more personalized features without constant internet connectivity.

• Google’s Pixel 9 Pro demonstrates successful on-device AI integration with photo editing capabilities and an advanced digital assistant
• Windows Copilot+ PCs currently face challenges in delivering their promised AI features
• The industry aims to eventually compress ChatGPT-level AI capabilities into mobile devices that can process data locally

Google’s Pixel 9 Pro highlights: The smartphone showcases several practical AI-powered features that operate directly on the device.

• “Add Me” feature allows users to insert themselves into group photos when they’re the photographer
• Audio enhancement tools can adjust individual speaker volumes and reduce background noise in videos
• Background replacement feature generates AI-created scenery, though results can appear slightly artificial
• Call Notes provides call recording and transcription, though the summary feature occasionally fails

Microsoft Recall’s mixed results: Microsoft‘s AI-powered memory tool shows promise but currently suffers from reliability issues.

• The software captures screenshots every few seconds and indexes both text and images
• Image analysis capabilities show potential but often misidentify content
• Many websites go unrecorded, limiting the tool’s effectiveness as a comprehensive memory aid
• Microsoft acknowledges the current version needs improvement

Privacy considerations: On-device AI processing offers enhanced security benefits compared to cloud-based alternatives.

• Microsoft Recall‘s data remains encrypted and stored locally
• Physical access to devices would be required to compromise AI-processed information
• Local processing reduces the need to share personal data with cloud servers

Looking ahead: The evolution of on-device AI points toward more sophisticated personal computing experiences, though current implementations vary significantly in their effectiveness and reliability.

• Early adoption challenges highlight the gap between AI’s potential and current technical limitations
• The industry continues working toward creating self-contained, personalized AI assistants that understand user preferences and anticipate needs
• Success will likely depend on balancing ambitious features with practical functionality