Google Swaps the Engine Inside AI Mode

Search Engine Journal reports that Google has made Gemini 4 Hydrogen the default model behind AI Mode for users worldwide, and that it now also handles the most complex questions in AI Overviews. For anyone who tracks how often AI answers mention their site, that is a major part of the system changing without warning.

The timing is the real story. According to the same report, the switch landed two days into Google's November core update, so two separate changes are now hitting search results at once. If your traffic shifts this month, you will struggle to tell which change caused it.

Why This Reshapes Search Rankings and AI SEO

Hydrogen is described as better at following instructions, handling messy or oddly worded questions, and breaking one question into several smaller searches at the same time. Google's stated aim is to surface more links to the web. That sounds encouraging for publishers, but it moves the goalposts rather than removing them.

Old-school SEO chased position one on a results page. AI SEO chases something different: being the source an AI answer decides to name. Those are not the same skill. A page can rank well and still never get cited, because the model tends to favour content that answers directly and can be checked quickly.

There is a measurement warning here too. Google has not said publicly which questions count as the "hardest" ones, so no one can separate that effect from the core update with real precision. Treat this month's numbers as a rough signal, not a verdict on your Google algorithm performance.

What This Means for Australian SMBs

Australian businesses are competing in the same AI layer as everyone else. A small retailer in Geelong or a consultancy in Perth can be named in an AI answer beside a national brand, because the model is choosing sources on relevance and clarity, not on ad budget.

That cuts both ways. It means genuine opportunity for well-written, specific content, but it also means less patience for thin pages that only exist to chase keywords. With tighter budgets than big competitors, small and mid-sized firms cannot afford to rebuild strategy every time Google changes models.

What You Can Do Now

  • Note the key dates in your analytics so you can compare traffic before and after the model switch and the core update separately.
  • Track citations and referral visits from AI answers, not just keyword positions. Rankings no longer tell the full story.
  • Write pages that answer real, messy, multi-part questions the way customers actually ask them.
  • Use clear headings, short answers, tables and plain summaries so a system can lift a clean answer from your page.
  • Hold off on big strategy changes for a few weeks. One month of noisy data is not a trend.

If you would like help making sense of it all, MS&VG's AI-powered SEO service is built to help Australian businesses stay visible as search keeps evolving.