THE TECH JOURNAL

A Human Take on AI, Tech, and Data.

Practical thoughts on how we work, build, create, and think.

The AI That May Never Break a Rule

AI can obey every narrow rule and still create harm at a much larger scale.

Useful AI completes tasks. Safe AI also needs to know when to refuse.

Some AI systems hit every target while quietly making the real problem worse.

Removing every delay and backup can make an organization faster and easier to break.

AI can ace the test by learning the wrong pattern, then fail when the world changes.

Search made us compare sources. AI makes it easy to accept one smooth answer.

AI saves mental effort, but some of that effort is how independent judgment grows.

The biggest runaway risk may be an AI system compromised by someone else.

The hard part is not finding the off switch. It is deciding who can use it.

AI alignment is not just technical. It is a question of whose values count.