Google's learning science team published a piece in the EdTech Insiders newsletter on how they build learning tools. It was written by Julia Wilkowski, who leads that team, and it was sponsored by Google. Both things are worth knowing. The principles are good. The way to read the numbers is the useful part.
The five principles
Google says it designs its learning tools around five ideas: encourage active learning, manage cognitive load, stimulate curiosity and motivation, adapt to the learner, and deepen metacognition, which means helping learners think about their own thinking. They draw on Piaget, Dewey, Vygotsky, Bandura and Papert. None of this is new and none of it belongs to Google. That is a good thing. It means any tool, from any company, can be held to the same list.
Good learning design principles are public. Hold every tool to them.
The study that stands out
Maths students in Sierra Leone who used Google's Guided Learning, with their teacher directing how it was used, gained the equivalent of 1.2 to 1.7 years of typical instruction on standardised tests within eight weeks. Over 90 percent of their interactions were step by step work, not requests for the answer. That second number is the interesting one. The gain came from a tool that made students work, used under a teacher's direction.
The number to read carefully
Teachers using Gemini for Education reported cutting time on administrative tasks by 70 percent. Reported is the key word. It is a self reported figure in a sponsored piece. It may well be true for many teachers. It is not the same kind of evidence as a test score.
The line we agree with most
"Edtech tools are most effective when teachers are in the lead." Wilkowski also names the worry directly: many teachers are frustrated about students using generative AI to offload their thinking. When the biggest player in the market says the teacher should lead, it is worth taking at its word, and checking whether the product actually works that way.
Four questions for any vendor's evidence
Who paid for the study, and who ran it? Was the result measured, or reported by users? Under what conditions did it work, for example with a teacher directing it? And does the product you are buying work the same way as the one in the study? Ask those four of Google, of us, of anyone.
Take these with you
- 1Good learning design principles are public. Hold every tool to them.
- 2The strongest result came with a teacher directing the tool and students working step by step.
- 3Self reported time savings are not the same as measured outcomes.
- 4Ask who paid, what was measured, under what conditions, and whether it is the same product.
