Day of AI Australia’s Jac Manison joined developers from around the world at Google’s I/O developer conference in San Francisco in May. In this week’s blog, Jac reflects on her main takeaways for educators.
It was difficult not to get caught up in the hype. The slick demos, the atmosphere, the crowd energy, the sheer scale of it all. But through every announcement, one thought kept pulling me back: what does this mean for Australian school students?
Being in the audience at Google I/O 2026, watching tech leaders from across the AI landscape share what they had built, what they were shipping, and where they believed things were heading, was genuinely extraordinary. Google I/O (short for “Innovation in the Open”) is Google’s annual developer conference where the company unveils its latest advances in software, hardware, and AI, setting the tone for where digital technology is heading. The announcements came fast: Gemini Omni, Gemini Spark, the Neural Expressive design overhaul, a reimagined Google Search, AI agents running silently in the background, smart glasses, and a quiet but unmistakable focus on physical AI and robotics.
But I’m a teacher, a Digital Technologies teacher who has spent years leading professional development as Head of Digital Learning. And we carry a particular kind of responsibility into spaces like this, to filter signal from spectacle, and bring something honest and useful back to classrooms, colleagues, and school leaders. So here is what I came away with, framed not for developers, but for educators.
Three takeaways for the classroom
Takeaway 1: Agents
Agents dominated the conversation at Google I/O. We saw similar emphasis on the growing role of agents at the Microsoft AI Tour, held recently in Sydney.
Gemini Spark, a persistent AI agent designed to run continuously, complete long-horizon tasks, and work alongside you rather than simply respond to you. Agents in Google Search: personalised AI systems that run in the background, watch topics on your behalf, and surface relevant actions at the right moment.
An example could be a student researching a history assignment setting up an agent to monitor new sources on their topic, say, “the causes of World War One”, and when a relevant article, video, or resource appears, the agent surfaces it automatically, ready for them to use.
Another example is a teacher setting up an agent to monitor for new curriculum updates or resources related to a unit they’re teaching, for example, “Digital Technologies Years 7–8”, and when the Australian Curriculum website or a trusted education body publishes something new, the agent flags it and suggests the relevant document or link, ready to review.
The agent’s action in this case is surfacing the information to you at the right moment: “Here’s a new resource that matches what you’re watching for.”
Some agents can go further and take more active actions, like drafting an email, adding an event to your calendar, or booking something, but in this example, the action is simply alerting the teacher to something relevant so they can decide what to do with it.
For education, this represents a shift in what learning could look like. Students could collaborate with AI agents that can plan, research, create, and complete tasks alongside them. The skill set we need to be teaching is evolving from “finding answers” to “working with intelligent systems.” Prompting, delegating, verifying, and critically evaluating agent output, these are the new literacies.
Takeaway 2: Multimodality
Gemini Omni, Google’s new model combining reasoning with advanced content creation, allows users to generate and edit video through combinations of text, images, audio, and voice prompts. You can describe what you want, upload footage, and refine it through natural conversation. Neural Expressive, the redesigned Gemini interface, replaces walls of text with interactive timelines, generated videos, and dynamic visual responses laid out in real time, personalised.
This signals a world where learning moves beyond typing and reading. Students can increasingly interact with AI through voice, images, video, gestures, and live environments, as naturally as they have a conversation. Text will remain important, but it will be one channel among many. As educators, we should start thinking about how they assess and develop student capability across these modalities, not just written output.
Takeaway 4: Interfaces
Perhaps the quietest but most far-reaching shift announced at I/O was in how we interact with software itself. Neural Expressive isn’t just a visual refresh, it signals that the “app” as we know it is changing. Google’s vision is one where you describe a goal in natural language, and the AI dynamically constructs the experience around you.
Traditional menus, toolbars, and step-by-step software workflows are being replaced by conversational, adaptive interfaces. For students who have grown up learning how to navigate specific apps, this could be seen as a disruption. However, for students, this is less of a disruption than it might seem, describing what you want, thinking through a problem, communicating with clarity, these are skills good teaching has always prioritised. The difference now is that the audience is an AI.
Across the Dialogues stage, sessions on robotics, including Google DeepMind and Boston Dynamics, pointed to something bigger than any software update. Physical AI: intelligence that doesn’t just process and respond, but perceives and acts in the world. Humanoid robots, embodied agents, AI-controlled physical systems. This was a dominant thread at I/O, and it is moving faster than most educators realise. The next generation of students will enter a workforce where physical AI is not science fiction, it is infrastructure. Schools could start conversations about this now: what roles look like, what skills matter, and what it means to work alongside systems that can do physical labour.
I left Google I/O with a sense of genuine excitement and an even stronger sense of purpose. Google launched several genuinely impressive tools at I/O, but they are far from alone. Anthropic, OpenAI, Microsoft and others are releasing innovative and potentially transformative tools with increasing frequency. What stood out to me across all of it is how these developments will force us to rethink how young people learn, how educators will need to adapt, and why AI literacy is becoming one of the most important things we can give students as these technologies make their way into our schools, homes, and workplaces. Education has always been at its best when it leans into the future rather than waiting for it to arrive. This is our moment. Teachers, school leaders, and students don’t need to wait for curriculum documents to catch up, we can start now, exploring, experimenting, and helping young people make sense of the world being built around them.
Day of AI Australia received funding from Google.org as part of the 2025 GenAI Accelerator to develop educational games. Day of AI Australia and UNSW Sydney team members were invited to Google I/O 2026 as members of the Accelerator Alumni community.
Bring AI literacy to your classroom with Day of AI Australia
Our AI literacy program gives students the knowledge and critical thinking skills to engage with AI more thoughtfully.
It’s free, curriculum-aligned, and designed to be taught without a computer science background.
