What’s your opinion on using AI for code review? Do these tools lead to meaningful code improvements, or do they add an unnecessary step to the code review process?

Currently an AI generated code once reviewed by another AI engine can still have findings, which means its not full proof currently. However, this should not discourage us for using AI for code review as an extra Step precursor to the main code review. This might lead to majority findings which might reduce the efforts of the reviewer before a final walk through.

It's a force multiplier. You can use it for code review and static security analysis, while at the same time saving work for your team. That's not to say that you should not put eyeballs on the code during review, but an AI analytics engine could make that process more productive and efficient.