EIRA

A reflective AI space to help society grow with, not against, AI.

Comment

Inspiration

We're in an age here AI has exploded, becoming a daily tool for nearly 78% of Indonesians. But this rapid adoption has created a new kind of divide. While some leverage AI to get ahead, many are being left behind, struggling with limited digital skills and unequal access to quality learning resources.

This creates a perfect storm. People are rightly worried about AI's potential to erode trust in information, amplify existing biases, and diminish critical thinking. Without proper guidance, AI can widen the equity gap and become a tool of manipulation.

That’s why we built EIRA (Ethical AI Reflection & Awareness). EIRA is dedicated platform to bridge the AI literacy gap in Indonesia. It’s designed to give everyone, regardless of their background, a fair chance to grow with, not against, AI by providing the knowledge and wisdom to use it safely and effectively.

What it does

EIRA is an all-in-one learning hub that makes mastering AI an interactive and empowering journey. Our platform are designed to build confidence and critical thinking from foundational concepts to advanced practical skills.

Here are EIRA's core features:

  • AI 101 - Literacy: Learn the basics of AI with our interactive, step-by-step approach. Each module is broken down into short videos, simple analogy cards, and quizzes based on real-world scenarios.

  • AI 101 - AI Platform Comparison: Compare the world's leading AI platforms (ChatGPT, Gemini, Claude, and other AI models) side-by-side to find the perfect partner for your specific needs.

  • AI 101 - AI Ethics & Policy Explainer: Jump into a strategic simulation where you play as a leader, making tough policy and ethical choices to see their impact on Indonesia's AI future.

  • AI 101 - AI Watch: Get the latest scoop on AI trends and issues in Indonesia. This section offers critical analysis and lessons learned from real case studies.

  • The Prompt Evaluator: Sometimes, you can't get the most out of an AI because your prompt isn't detailed enough or might contain hidden bias. This tool helps you analyze and improve your prompts to get better, more accurate, and fairer results.

How we built it

Frontend: React, TailwindCSS, and Firebase

Backend: Flask, FastAPI, Pytorch, HuggingFace, and Google Gemini API

Challenges we ran into

  • Understanding the specific context, policies, and user behaviors surrounding AI utilization in Indonesia, which required further research

  • Navigating the complexities of the deployment environment

  • Addressing the potential for bias when using one AI model to evaluate another

  • Working within tight time constraints in a fast-paced environment

Accomplishments that we're proud of

  • Implementing prompt engineering best practices for Large Language Models (LLMs) to achieve highly promising results.

  • Successfully deploying and utilizing a local Large Language Model (LLM) from GoTo: GoToCompany/gemma2-9b-cpt-sahabatai-v1-instruct

  • Creating polished and engaging animations using TailwindCSS

What we learned

  • The complexities of AI model deployment and evaluation, particularly for LLMs and their application in assessing human-generated content

  • How to utilize AI models to their full potential within the software development lifecycle.

  • Proper prompt engineering techniques, including how to parse model outputs to achieve the desired results

  • The importance of teamwork, seamless front-end/back-end integration, effective time management, and clear communication

  • How to deploy an LLM in a GPU-based environment (Google Colab) using the ngrok service for external access

What's next for EIRA

While EIRA's launch is timely amidst the rapid growth of AI and the clear need for public education, we recognize there is always room for improvement. Looking ahead, here are several key areas we aim to develop to further enhance inclusive AI education:

  • Stakeholder Engagement: To foster greater involvement from stakeholders (government, industry, and academia). This will help ensure a balanced approach where AI is properly regulated while its full potential is harnessed for the public good.

  • Integration into Education: To pursue the deeper integration of EIRA into the formal education system, particularly within Indonesia.

  • Native Applications: To develop native applications for both web and mobile platforms, improving accessibility and providing a smoother user experience.

  • Resources for LLM Exploration: We are aware that developing Large Language Models (LLMs) is costly and that their vast capabilities remain largely unexplored. Therefore, securing the right environment and sufficient computational resources will be crucial to support broader user access and continued innovation.

Built With

Share this project:

Updates