Fearless Schools in AI for Education: A New Trust
Artificial intelligence in education is no longer a distant vision of the future. It’s a tool with enormous potential to transform learning. However, with great power comes great responsibility. What must be done to ensure that technology genuinely helps, rather than harms? How can we be certain of its accuracy, security, and fairness?
At BlueRider.Software, we believe that trust in AI must be earned. Our case study is Assistik.ai, which we are developing together with the Fundacja Szkolna (School Foundation) for Kasai sp. z o.o. We will show how technology and responsibility work hand in hand.
When it comes to developing educational technology, there is no room for error, especially with AI. Google Gemini was chosen as the heart of Assistik.ai, as it fits perfectly into the environment where our partner, the Fundacja Szkolna, operates. We are aware that we cannot blindly trust artificial intelligence. That’s why our experts conduct a series of demanding tests, checking even the smallest details.
We create scenario-based tests that verify the AI’s responses:
- Are the answers precise?
- Are there any factual errors?
A digital shield for student data: Application Security
The most important point is user security – in this case, the students. Data protection is our absolute priority. We package our application in Docker containers, and every update goes through an automated scanner that searches for known vulnerabilities and system weaknesses. This ensures that the system we are building is secure.
AI Without Bias? Our Ethical Audit in Practice
Technology in education should be objective to fulfill its purpose. So, where can we find the right model to follow? Thanks to our subject-matter partner, the Fundacja Szkolna, this wasn’t an issue. We exclusively use educational materials provided by them.
Their knowledge is the foundation of Assistik.ai. The algorithm we created is checked against the materials supplied by our partner. This helps us ensure that the generated questions are not only factually correct but also consistent with the spirit and mission of the Fundacja Szkolna. Our goal is to create a virtual, personal teaching assistant that helps students achieve their own goals.
Questions for the Expert: Ethical AI in Practice
1. What happens when the AI makes a mistake?
Our verification process is one of continuous improvement. Every response created by the AI is instantly logged and saved in our database for later analysis.
Furthermore, we’ve implemented an advanced two-step verification system. You can think of it as a collaboration between two specialized AI “agents”:
- The Generator-Agent: Creates questions based on the materials provided by the Fundacja Szkolna.
- The Verifier-Agent:Operating on a more critical set of prompts, it reviews the questions generated by the first agent. It examines them from various angles and automatically rejects those that do not meet our rigorous quality standards.
We save these verification results as well. This way, we not only eliminate weak responses immediately but also collect data that allows us to continuously refine the entire system.
2. How do we balance quality control with Agile speed?
To reconcile comprehensive quality control with the Agile methodology, we rely on metrics and full automation. Our system evaluates every dataset three times, averaging the results to minimize subjectivity and provide a reliable overview of the model’s effectiveness.
The entire process, including report generation, is automated. This gives us the ability to rapidly assess the AI’s effectiveness while maintaining the full dynamics and speed of our workflow.
More than code: Our work philosophy in practice
Behind every good product, there is always the right process and partnership. At BlueRider.Software, our approach is based on transparency, agility, and discipline. These three pillars allow us to create solutions that meet client requirements.
Agile: Building shoulder to shoulder with the client
Assistik.ai is an example of our agility in practice. Working with the project owner, Kasai sp. z o.o., and our partner, the Fundacja Szkolna, we didn’t expect a complete vision at the outset. The Agile methodology is based on a step-by-step approach. Throughout the project, we are in constant dialogue with the client and our subject-matter partner, which helps us shape it together. Both the Fundacja Szkolna and Kasai have become true partners in the creative process.
DevOps: Automated quality control
Our approach is supported by solid technology. The DevOps philosophy significantly shortens software production time by automatically checking our code – even the smallest change is automatically tested. Systems check its readability, standards, and potential errors. This yields better results in less time, perfectly aligning with the Agile methodology. All of this happens before anyone even runs the application.
Summary: Responsibility is Key to Safe Innovation in AI for Education
Assistik.ai shows that in today’s world, people’s trust is more important than the technology itself. Our approach is based on three pillars: accuracy, security, and ethics. At BlueRider.Software, we rely on Agile and DevOps methodologies. This allows us to step into our client’s world. Using the Google ecosystem, which the client was already using, was a natural choice to create a cohesive and secure tool. This is what modern and responsible software development looks like to us.
Want to know more? Check out these links
If the topics we’ve discussed have interested you, take a look at these resources. They will help you expand your knowledge.
Our resources – BlueRider.Software:
Learn more about our company and work philosophy
See how we can help with your AI project
Have an idea for a project? Let’s talk about it
External links to broaden your knowledge:
Responsible AI at Google: Learn about the principles that guide Google in creating ethical AI. Learn more about Responsible AI at Google
Security in Docker: Official documentation and best practices. Read about Docker Security
Guide to DevOps: Resources from Atlassian explaining what DevOps is. See the Atlassian DevOps guide
AI Ethics in Education: A government publication on the challenges of AI in Polish schools. Read the guide “What AI is not for”