1Edtech Why AI in Learning Needs Standards: Shared Context, Shared Trust , Westminster Hall.
It was really useful to get high level OECD view on Monday and technical deep dive from 1Edtech on following day. Open technical standards still drive most of the global developments in technology and learning.
Great presentations and discussion showing how AI can bridge different educational lexicons when it has been trained. The challenge is being confident enough to start this process.
- Japan, Korea and France identified as world leaders in adoption of open standards - in terms of national data.
- Korea - using AI already to map national competency standards to national qualifications. ( accelerates development process)
- Japan - Using the one roster standard to streamline the exchange of educational data between School Information Systems (SIS) and Learning Management Systems (LMS) across the country.
- France - Whole sale adoption of open standards to improve learning and data flows across their educational system.
The discussion and debate was unavoidably technical. A lot focussing on what needs to be done to pre-existing content and data to give it context to work well with AI systems.
Useful metaphor - example - the same ingredients can make a Pancake or a Yorkshire pudding. Currently a large language model with no context would not know which one you want.
When you have context. You can do more
You can use an MCP Server - MCP servers are essential for building AI agents that can actually "do" things, rather than just talk about them. It allows multi agentic workflows. MCP is a protocol and a standard , it acts as a "USB-C" connector for AI, allowing models to securely access local files, databases, and third-party services without requiring custom, one-off integrations.
These systems use CASE
- A data standard that allows educational institutions, employers, and edtech platforms to exchange competency data in machine-readable formats (JSON).
- Purpose: It replaces static, inefficient documents like PDFs or spreadsheets with standardized, digital frameworks that can be easily mapped, searched, and updated.
- Frameworks: Digital versions of academic standards or professional competencies.
- Items: Machine-readable statements of what a learner should know or be able to do.
- GUIDs (Globally Unique Identifiers):Every standard, skill, or competency is given a unique identifier, ensuring it can be recognized globally across different systems.
- Associations: Links that define how competencies relate to each other, such as aligning a local school curriculum with national or international standard.
- In the context of global learning (interconnected education,, sustainability, and international skills), CASE serves several critical functions:
- Interoperability for Global Education: As educational institutions operate internationally, CASE allows them to map their curricula, learning outcomes, and assessment rubrics to common global frameworks, facilitating easier student transfer and recognition of skills.
- Workforce Competency Mapping: CASE helps define and exchange skills needed for the global economy. It is used in higher education and corporate training to ensure that digital credentials and competency-based learning modules are recognized across different borders and platforms.
- Digital Credentialing: It enables the alignment of learning outcomes with micro-credentials and digital badges, allowing a skill earned in one country to be verified and understood by an employer in another.
- Reducing Inefficiency: By providing a common, machine-readable language, CASE eliminates the need for manual, error-prone, or time-consuming mapping of learning standards across different systems.
Example used Campus Mind - system can identify where student potentially needs some additional support and can build a personalised course from materials that institution already has across range of repositories including the virtual learning environment.
Microsoft and Google are both building integrations. Jisc standards group moving ahead in UK.
- Otus - taking data and learning materials from multiple sources.
- French Teacher training Moodle platform - AI was used to - re-engineered with standards and context to allow individual training pathways - personalised for teachers. The MCP server can now be used with any learner management system.
- Call to move away from general models of AI to local models of AI that capture context etc ( what OECD said)
- To move forward all players need increased awareness of the open standards that are already available and ones in development.
- We really need SQA - Qualifications Scotland to get a move on here. This is much more than end of paper based certification.
- More Scottish organisations should have appropriate membership of 1Edtech and I am going to see if I can help with that.
- And a final plug anyone interested in these open standards should get along and or get a paper into their upcoming European Conference in Salonika.
- AI and MCP services can start bringing your data together securely. You will be using local engine rather than opening up anything. Get a move on.
- Make sure you and or any suppliers you use are making most of open standards. Some of this should be hard wired into procurement guidance.
No comments:
Post a Comment