Google Just Made Talking to AI Feel Less Weird (September 16, 2026)

You know that stiff feeling when you talk to an AI on your phone?

You pause. It pauses. You talk over each other. It starts answering a question you already moved past. Half the time you hang up and type instead.

Google just shipped an update aimed straight at that problem.

On September 15, Google introduced two new voice models:Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking. The short version:they want talking to Gemini to feel more like talking to a person who can keep up.

Here's what that means in plain English.

Gemini 3.8 Live is built for fast, back-and-forth conversation. It can look at what you're pointing the camera at while you talk. It can switch between languages mid-chat. Google says it supports 97 languages and can change mid-sentence without you starting over. It can also kick off a tool or lookup in the background and keep talking to you while that finishes.

The Extended Thinking version is for harder jobs. Google says it can reason and speak at the same time. You get those little human cues like “let me check that,” then a live walkthrough while it works through steps. Same idea as a coworker who narrates what they're doing instead of going silent for thirty seconds.

Where can you actually try this?

According to Google's own post:

• 3.8 Live is rolling out in Search Live for everyone.
• 3.8 Live Extended Thinking is rolling out in Gemini Live.
• If you pay for Google AI, pieces of this are also landing in Workspace tools like Docs, Gmail, and Keep, with different access by plan.
• Developers can hit the Gemini API and Google AI Studio.

One detail worth knowing if you care about deepfakes:Google says all audio from these models carries a SynthID watermark. You can't hear it. The point is that the audio can still be checked later as AI-made.

Look, I get it. Voice AI has been a party trick for a while. Cool demo. Clunky in real life. This is Google saying the everyday stuff (Search, the Gemini app, your inbox) is where the next fight is.

If you already use Gemini on your phone, open Live and see if the conversation feels less brittle than last month. If you don't, Search Live is the lowest-friction test. Ask it something you'd normally type. Interrupt it once on purpose. See if it recovers like a person would.

Does talking to AI ever feel natural for you, or do you still bounce back to the keyboard?

Source:Google's announcement of Gemini 3.8 Live and 3.8 Live Extended Thinking (September 15, 2026).

Written by

Joe Foley

Joe Foley is the creator of AI for Ordinary People. He helps beginners, parents, creators, and small business owners use AI in simple, practical ways. Joe has been podcasting since 2013 and creates plain-English guides, prompts, and workflows for people who want useful help without tech jargon. His goal is to make AI feel less confusing and more useful for real life, real work, and everyday decisions.