A lot of the AI conversation still sounds like it lives in tech companies and venture capital decks.
But the signals are getting clearer. Education and workforce development are moving to the centre of national AI strategy. Not as a side topic. As the main way countries try to stay competitive, keep people employed, and avoid new divides opening up.
This week's reading landed in three layers for me:
- governments are treating skills and literacy as strategic infrastructure
- the public is using AI more, and using it differently than last year
- teachers are already using AI at scale, but the impact depends on how we test and support it
Governments are starting to treat AI skills as infrastructure
The State of AI Report 2025 highlights how education and workforce development are becoming central to national AI strategies.
It points to governments investing in skills, training, and literacy so people can actually take part in an AI-driven economy, not just watch it happen.
That makes sense to me. If AI becomes a general-purpose capability like the internet, then "AI skills" are not optional extras. They are part of the operating system of a country.
Link: Read the State of AI Report 2025
People are using AI a lot more and it is shifting away from content creation
A new Reuters Institute report shows weekly use of generative AI tools jumped from 18% to 34% across six countries in the past year.
ChatGPT is still dominant, but embedded tools like Gemini and Copilot are pushing exposure wider, because they appear inside tools people already use.
What's changed most is how people use AI.
Information-seeking has overtaken content creation as the top use case. People are turning to AI for research, facts, and advice. Stuff that used to belong to search engines and news outlets.
For publishers, the findings are sobering:
- 61% of Americans have seen AI-generated search answers in the past week
- only one-third regularly click through to original sources
- trust in AI summaries is moderate but conditional
- just 12% are comfortable with news made entirely by AI
The public sees value in AI for editing or translation, but still prefers journalism led by humans, supported by technology.
This matters for schools because it changes student behaviour. If AI becomes the "first stop" for information, then AI literacy needs to include source checking, judgement, and healthy scepticism, not just "how to prompt".
Link: Read the full report: Generative AI and news report 2025
Australian teachers are already heavy users and that is a big deal
The Conversation summarised new data from the OECD's Teaching and Learning International Survey (TALIS).
A few points stood out:
- 66% of lower secondary teachers in Australia used AI in the past year
- that's the fourth highest rate in the OECD and well above the 36% average
- most use it to brainstorm lesson plans and summarise content
- few use it for student assessment or data analysis, showing caution around privacy and professional judgement
This lines up with what I see. Teachers aren't waiting for some future "AI moment". They're already using it, and mostly in the places where it reduces workload without touching high-stakes judgement.
That's a useful reality check for anyone still asking whether AI for schools is "coming". It's here. The real work is making sure it's safe, fair, and actually improves learning rather than just speeding up admin.
Link: Read more: Australian teachers are some of the highest users of AI in classrooms
Why I built CurricuLLM as AI for schools
CurricuLLM is an AI built for schools and it's free for every teacher in Australia and New Zealand.
Over the past few months, in my spare time, I've poured my energy into building an AI made for schools. What I've learnt is that curriculum is at the core of a great AI experience.
Teachers should be able to trust that answers are safe, relevant, and matched to what students are meant to learn.
Teachers can use it to:
- save time preparing materials for lessons
- draft rubrics, assessments, and reports aligned to outcomes
- personalise activities without creating everything from scratch
From the start, my focus has been equality of access. That's why there will always be a free tier with fair-use limits. For those who want more, there's a small monthly option to remove restrictions.
This is just the beginning. I'll be bringing more states and curricula online over the next few months as licensing arrangements are finalised.
Link: Visit CurricuLLM
Personal AI devices could make AI even more ambient
There's also an interesting thread about OpenAI and Jony Ive working on a new AI-powered device.
The goal is a palm-sized personal assistant that understands the world through voice and vision, without relying on a screen.
Reports suggest challenges around software, privacy, compute needs, and even the device "personality", how much it should speak or listen.
If devices like this land, it will shift expectations again. AI won't just be something you open in a browser. It will sit closer to daily life.
And again, that loops back to schools. The more ambient AI becomes, the more important it is that students learn how to think with it, not just use it.
Link: Read more: OpenAI and Jony Ive working on new AI-powered device (FT)
A needed reminder from Justin Reich about humility and evidence
Justin Reich from MIT has a piece that I think every AI-in-education leader should read.
His point is simple. Educators have seen "transformative" tech before. The tool rarely delivers by itself. Outcomes depend on how schools use it, and durable gains come from steady, evidence-based experimentation.
His takeaways for AI in education:
- be humble, today's approaches are early guesses
- experiment carefully, some parts of the curriculum can test AI and some need protection
- assess locally, compare work before and after AI use to measure real impact on learning
I agree with this. AI may become as essential as the web, but the outcome depends on how thoughtfully educators test, adapt, and share what they learn.
Link: Read more: What past education technology failures can teach us about AI in schools
Where I'm sitting after all this
Education is now part of national AI strategy because it has to be.
Public AI use is rising fast, and it's moving into information-seeking, which changes how students find, trust, and use knowledge.
Teachers in Australia are already among the most active users in the world, but use is still concentrated in planning and summarising, not high-stakes assessment.
So the next step for AI for schools is not "more tools".
It's better support, better evidence, and better design:
- tools that are grounded in curriculum and progression
- professional learning that helps teachers build confidence and judgement
- careful experimentation with real measurement, not hype
- clear expectations for students about integrity, sources, and thinking
If we do that, AI can lift learning and reduce workload.
If we don't, we'll still get adoption, but it will be uneven, and the gaps will grow.