As AI gets better, it reveals an empty promise

Google's new Gemini AI agent, called Spark, is turning heads this week after hands-on testing by The Verge's David Pierce and Jay Peters. Both came away with the same reaction: it's effective enough to be unsettling.

Spark demonstrated an uncanny ability to infer personal information without being explicitly told. It knew David's dog is named Frida and recalled Jay's wife's first name, even though neither journalist directly provided these details to Google. The AI pieced together context from their digital footprints.

But the real concern isn't just about privacy. It's about what we're optimizing for. These tools are being pitched as productivity solutions, as if being more efficient at email and task management will somehow fix the deeper structural problems in how we work.

The productivity framing often carries a moral undertone, suggesting that if you're struggling, you just need better tools or better habits. But that misses the point entirely. No amount of AI assistance will address burnout, unrealistic workloads, or broken workplace cultures.

For AI users and professionals, this is worth thinking about. These tools can genuinely help with specific tasks, but they're not a substitute for questioning whether the work itself makes sense. The risk is that we automate our way into doing more of the wrong things, faster.