Ask HN: Is Google NERFing AI Mode?

4 points·PaulHoule··4 comments

Sorry to ask a question about AI, but... Lately I've noticed that Google seems to be hiding the AI mode and trying to push people into the less capable AI Overviews. Often when I go explicitly to the AI Mode form it kicks me to AI Overview and doesn't give me a chance to converse more -- are token costs catching up with them?

View on HN

not sure, hard to judge UX because they are always changing and serving different version to different people, but the ai overviews and ai mode are why I stopped using google, they have always been very bad i.m.e.

I leave them feedback everytime they shove that in my face. I've mostly stopped using Google at this point and use a custom search setup (exa, tavily, searxng) with an agent (optionally) in front of it

Wonder if the German ruling that they are liable for falsehoods in overviews is changing this behavior?

My impression is that the AI overview sucks and the AI mode is pretty good.

Often they give diametrically different answers to questions and in those cases the AI mode is usually right.

It can be way better with your own "deep research" agent that does not insist on writing a half baked report, the main idea is that they can make multiple searches and sift through results. It can be instructed to be very concise or to write a long ass report, your choice.

This is my custom MD kb tool with this feature

https://github.com/verdverm/gmd

The redirect from AI Mode to AI Overview without follow-up feels like cost management—conversational queries burn way more tokens than one-shot summaries.

Reply on news.ycombinator.com