Navigating the ‘messy middle’ of AI in EdTech

Each year, the ASU+GSV Summit offers a front-row seat to the fast-evolving intersection of education, technology, and the future of work. Our Director of Ufi Ventures, Helen Gironi, headed to San Diego earlier this month to attend the AI Show and Summit, joining thousands of education and workforce learning leaders.

At the 2025 events, one thing was clear: artificial intelligence (AI) is no longer just a buzzword, it is reshaping how we teach, learn, and work. While the promise of AI is exciting, we are still navigating the ‘messy middle’ of adoption, where experimentation is high, but long-term integration is still a work in progress.

Here are our key takeaways from this year’s ASU+GSV:

Best use cases: Embedding AI into everyday learning and work

The most successful applications of AI are not flashy, they are embedded into existing workflows, making everyday tasks faster and more impactful:

• Feedback at scale: AI integrated into platforms like LMSs has increased teacher feedback tenfold, allowing for quicker, more consistent support for learners.
• Conversational AI: Moving beyond screens, AI is being used to create dialogue and build community, particularly valuable in remote or hybrid settings.
• Career guidance: From retaining institutional knowledge to teaching students how to prompt AI effectively, these tools are expanding access to career pathways and networks.
• Economic mobility: AI skills are proving a gateway to roles offering £30k+ salaries and upward career potential.

Interestingly, many innovations are being driven by administrative staff, those closest to the flow of work and the practical realities of learners’ day-to-day experiences.

The Challenges: Skills, access and resilience

While AI offers huge potential, the challenges are significant:

• Teacher support: Effective use requires understanding educators’ digital confidence and providing hands-on help, not just one-off training.
• Learning gaps: Students may lean too heavily on AI, skipping foundational steps (e.g. ask AI to build a financial model, but them not knowing how it works).
• Digital divide: AI risks widening existing inequalities and economic gaps. Making tools accessible and inclusive is essential.
• Feedback fatigue: More frequent AI-driven feedback has, in some cases, required teachers to do extra lessons on resilience because pupils found the extent of feedback hard to take.

Institutions are also grappling with concerns about data quality, bias, privacy, and ensuring that AI adoption supports, but does not replace, human relationships.

AI transformation: In the ‘messy middle’

We are past the AI ‘big bang’, but most organisations have not yet embedded AI into daily operations. This in-between phase is where foundations are built:

• Slow uptake in education: Despite growing awareness, only 2% of US teachers were using AI regularly by late 2024, unchanged from a year earlier.
• Changing job landscape: AI-driven job displacement, especially in gig roles, is already underway. Yet new jobs are also being created, though less visible for now.
• Key skills for the AI era: Providing skills such as management to entry level workers and the ability to work hands-on with AI.
• Learner agency: Institutions like Michigan University are giving each student an AI coach to help them navigate their own learning and career choices.

As AI technology continues to accelerate, the future of work will not wait. With the right investment in AI-driven EdTech and VocTech, delivered ethically, accessibly, and intelligently, we can shape a workforce that is resilient, adaptable, and ready for whatever comes next.

To keep up with the latest insights from Ufi Ventures and to join the conversation on the future of learning, AI, and work, follow us on LinkedIn and Bluesky.