This year went by quickly! 2025 brought another onslaught of attention to artificial intelligence. Education had its own global issues of funding, policy, staffing, and more. AI added complexities and opportunities that are still being made sense of.
The one thing I know as I reflect on this year is that we haven’t yet harnessed all of the opportunity that AI has to redesign (and dare I say, recreate) our educational systems.
So what are the highlights from the last few weeks of the year? Here is what hit the top of my list.
- Many of you have heard me talk about Manus previously. It’s a great tool to just explore the possibilities in how content can be organized and shared. We have used it at home when our high school student created multi-modal study guides using their own notes (see an example here). This week Meta announced the purchase of Manus. This will be one to watch as things unfold.
- Digital Promise shared a new report, What States Say about Evaluating AI in Education: Reviewing Guidance from 32 States and Puerto Rico. The findings suggest that most states focus on defining terms, gathering stakeholder feedback, and establishing guardrails, while fewer are conducting pilot studies or rigorous outcome-based evaluations. In addition, the report highlights the seemingly growing alignment around using co-design and structured feedback loops that elevate educator, student, family, and community voices.
- Throughout the year the Stanford Institute for Human-centered AI examined AI’s impact on childhood safety, workplace dignity, mental health care, and privacy rights. They recently published a year-end roundup of its most-read stories, reflecting the central AI questions and concerns that captured public and policy attention in 2025. Readers were drawn to analyses that treat AI not as abstract potential but as technology with tangible, nuanced effects on everyday life and institutions. The top-read story examined risks in AI therapy chatbots. Stanford HAI research showed that popular chatbots sometimes reinforce stigma and can respond dangerously in suicidal intent scenarios, underscoring the need for human oversight in clinical and support settings. The 2025 AI Index featured prominently, documenting rapid cost declines, smaller models matching larger ones on performance, widespread adoption in business, and an increase in AI-related incidents. Another key story focused on emerging threats to children’s safety, notably AI-enabled “undress” applications that easily create synthetic explicit images. Research on worker expectations highlighted a disconnect between tasks workers want AI to support and what organizations prioritize automating. This is often seen in the work we are actively doing with schools, districts, and organizations. The common threads among their most read articles depict growing public interest in AI evaluation, safety, governance, and how AI intersects with human systems and social norms. Things we need more attention to in education.
- Last week, ADQ and the Gates Foundation announced an AI-for-Education partnership that includes an EdTech and AI Fund aimed at scaling evidence-backed learning interventions at national level, with a strong focus on sub-Saharan Africa. This is worth watching if you build tools meant to move from pilots to system rollout. This is amazing news for the Ai-for-education.orginitiative at Fab-AI. The team has continued to be forward thinking connectors in Africa and their ability to mesh technical complexities with community building is something to watch.
- Just before the end of the year, Japan approved its first basic national plan for AI, emphasizing “reliable AI” and accelerated adoption across society. For education and training teams, it is another indicator that national AI strategies are moving from principles to implementation expectations.
- The Inter-American Development Bank highlighted a national pilot in The Bahamas using the Edutec Guide, including teacher and leader self-assessment and practical work with AI for lesson planning with a digital citizenship lens. If you build PD or readiness tools, the design (diagnose, reflect, plan, resource) is the takeaway.
Looking back on 2025 here are some lessons learned.
As 2025 comes to a close, a few clear lessons stand out from the work, conversations, pilots, and policies shaping AI and learning this year. None of them are especially flashy. All of them matter.
AI readiness is less about tools and more about clarity of purpose.
Across schools, nonprofit organizations, and systems, the strongest work this year started with a clear problem of practice. Where AI was introduced to solve real problems of practice adoption was more thoughtful and impact more measurable. Where AI showed up as a solution in search of a problem, it often stalled or quietly disappeared.
Educators are not resistant to AI. They are resistant to uncertainty.
What we saw repeatedly is that teachers and leaders want guardrails, examples, and permission to experiment safely. Clear policies, shared language, and concrete use cases reduced fear far more effectively than inspirational messaging ever could.
Responsible AI lives in workflows, not documents.
Ethics statements and policy PDFs are necessary but insufficient. The most credible approaches embedded responsible use into everyday practice. Tool selection processes, data handling routines, assignment design, and review cycles all carried ethical intent forward in ways that static guidance could not.
Equity is shaped by design decisions, not intentions.
Low bandwidth access, language support, disability inclusion, and culturally relevant examples were the difference between AI expanding opportunity or reinforcing existing gaps. The year reinforced a simple truth. If equity is not designed in at the start, it rarely appears later. I am excited to share more of what I have been thinking about here during my Keynote at the CEC International Conference in March.
Assessment practices are under real pressure.
AI did not “break” assessment, but it exposed how fragile many existing models already were. Systems that leaned into performance tasks, process documentation, and reflective practice found more stability than those trying to preserve paper-based or surveillance-heavy approaches. With our Seedlings to Scale project fully funded by IES, the work we are doing at CoGrader is showing promising results in supporting educators globally.
As we head into 2026, the signal is clear. The question is no longer whether AI belongs in learning environments. The question is whether we are willing to do the slower, more disciplined work of aligning AI with what we actually value about teaching, learning, and human development.
Systems readiness has been top of mind for me this year and I am hopeful that educators and leaders lean in during 2026 to be sure that their systems are prepared for the power of AI to support or complicate their practice.
So what is up for 2026?
Thinking about what I have written about in 2025 here is what I think is on deck for 2026.
AI readiness will replace AI adoption as the dominant framing. In 2026, more education systems will stop asking “Which tools should we use?” and instead focus on organizational readiness. Expect greater emphasis on governance, staff capacity, data practices, and instructional coherence before tools ever enter classrooms.
Assignment and assessment redesign will accelerate, unevenly. I fairly consistently pointed to assessment as the pressure point this year. In 2026, systems that already started redesigning toward process-based, performance-based, and reflective assessment will gain momentum. Others will double down on restrictive or paper-based approaches, creating widening gaps between instructional models. One thing I don’t want to see is a resurfacing of the “blue book” — the old school paper based assessment that too many think is “AI proof” (news flash: that’s the wrong way to address AI in any classroom).
AI literacy will split into distinct competency tracks. A single definition of AI literacy will no longer hold. Educators, students, leaders, and developers will each require differentiated competencies. 2026 will surface clearer role-based expectations rather than one-size-fits-all frameworks.
Responsible AI will move from policy to operational practice. The articles repeatedly argue that ethics must live in workflows. In 2026, credibility will come from documented processes such as tool review cycles, data handling routines, and instructional decision protocols, not from standalone policy statements.
Low-resource and global contexts will shape innovation, not follow it. A consistent theme is that constraint drives clarity. In 2026, approaches developed for low-bandwidth, multilingual, and under-resourced settings will increasingly inform mainstream AI design and professional learning models.
Educator trust will become a more predominant indicator of AI success. My writings this year suggest resistance is rooted in uncertainty, not opposition. In 2026, systems that invest in shared language, safe experimentation, and transparent decision-making will see stronger uptake than those relying on mandates or enthusiasm alone.
AI governance will become a cross-functional responsibility. Ownership of AI decisions will shift away from isolated technology teams. Expect broader involvement from curriculum leaders, special education, legal, and assessment teams as AI touches more core functions. This is continually coming up as part of our work at the AI and Education Studio and we only expect it to deepen.
Equity claims will be scrutinized more closely. I 100% think that 2026 will bring sharper questions about who benefits from AI integration and who does not. Design choices around accessibility, language, disability inclusion, and data representation will increasingly determine whether initiatives are sustained or challenged.
Pilot fatigue will force fewer, better experiments. I posed frequent critiques of endless pilots. In 2026, leaders will demand clearer success criteria, exit plans, and pathways to scale. Experiments without a growth strategy will struggle to survive. In addition, the app generation may give way to platform growth where projects, custom GPTs, and GEMs in Gemini can do more than one off apps.
AI conversations will shift from fear to professional identity. Rather than asking whether AI should be used, educators will increasingly ask what it means to teach, assess, and mentor well in AI-supported environments. In 2026 I hope to see this reframing shape professional learning and leadership conversations at a deeper level.
And that’s a wrap on 2025. I hope you all have an amazing celebration in the days ahead and reflect on all of your success in 2025 while planning for 2026. One thing is for sure — AI isn’t going anywhere so hold on, it’s about to get real.
Remember that the future of education is being shaped now and we need more voices that are informed, thoughtful, and equity-driven. We need informed, thoughtful voices to make meaningful choices. Subscribe to Jody’s newsletter for weekly updates as soon as they’re published at https://jodybritten.medium.com/subscribe
