Google I/O 2025 highlights: Gemini powers smarter Search, personal AI; concern for web publishers

Google on Tuesday unveiled a sweeping array of artificial intelligence advancements, immersive video communications, and personalised digital experiences that are changing the way people get information and curtailing the flow of internet traffic to websites.

At its annual I/O developers conference, the tech giant launched “AI Mode,” a new feature that transforms how users interact with Google Search, shifting from traditional keyword queries to fluid, expert-like conversations.

The feature is now rolling out to all users in the United States, just two and a half months after limited testing began in its Labs division.

“We are in a new phase of the AI platform shift, where decades of research are now becoming reality for people all over the world,” said CEO Sundar Pichai to a packed crowd near the company’s headquarters in Mountain View, California.

Search gets smarter, but traffic to websites slips

The expansion builds on Google’s integration of conversational AI summaries, known as AI Overviews, which have become a prominent fixture atop results pages since debuting last year.

Used by 1.5 billion people globally, the tool has dramatically changed search behaviour, prompting longer, more complex queries and fewer traditional clicks.

But while Google claims the changes would trigger more searches and ultimately more clicks to other sites, external data tells a different story.

A recent study by search optimisation firm BrightEdge reported a 30% drop in clickthrough rates from Google’s results pages, raising concerns among web publishers about declining traffic and visibility.

According to the Associated Press, Google's own AI mode acknowledged that the company's search engine seems unlikely to be significantly hurt by the shift to AI technology. The feature warned that web publishers should be concerned about AI mode reducing the traffic that they get from search results.

AI-powered agents and future tools

Google’s foray into digital agents took centre stage with updates to Project Mariner, its early AI assistant prototype. The company announced plans to begin testing agents capable of booking tickets, making restaurant reservations, and conducting searches via live video feeds, pointing toward a more proactive and hands-free search experience.

Google Beam and real-time translation

Google also introduced Google Beam, the next evolution of its Project Starline 3D video technology. Beam uses AI and an array of six cameras to transform standard 2D video calls into realistic, three-dimensional conversations in real time. Early versions will ship to select customers later this year, in partnership with HP.

Speech translation is also coming to Google Meet, allowing real-time translated conversations with matched voice tone and facial expressions. English-Spanish beta support begins rolling out now, with more languages coming soon.

Personalisation push

A major theme this year was personal context, using user-approved data from Gmail, Google Drive and other services to personalise AI assistance.

Gemini can now generate Smart Replies in Gmail that mirror a user’s tone, style and even favourite phrases based on past conversations and documents.

New media tools: Imagen 4, Veo 3 and Flow

Google introduced Imagen 4 for high-fidelity image generation and Veo 3, its latest video model with native audio generation. Creators can now use these tools within the Gemini app, alongside a new cinematic editing tool called Flow, which turns short clips into full scenes, opening fresh creative pathways for filmmakers and content creators.