LLM Capabilities

AI systems are rapidly evolving beyond simple chat interfaces. It is important to know what they can do and how to use them effectively for learning.

💡 What Can LLMs Do?

LLMs can be a powerful partner in your data analysis journey. With the right prompts, they can:

• 🧠 Explain unfamiliar concepts – Break down Pandas and Matplotlib syntax and data analysis concepts
  
• 🐞 Debug and optimize – Suggest debugging strategies and code optimizations
  
• 🧪 Provide examples – Show how to use libraries like pandas, matplotlib, and numpy for data manipulation and visualization
  
• 🔄 Refactor and document – Improve your code structure and add clear documentation
  
• ✨ Generate code – Create code based on text descriptions (i.e. prompts)
  
• ✍️ Writing and summarization – Generate reports and concise summaries of your analysis
  
• 🌐 Translation – Convert text between languages for international datasets
  
• 💡 Idea generation – Provide creative suggestions for analysis approaches
  
• 📈 Analyse data – Can help analyse data from either text or visual input

They are especially useful for:

• 📊 Exploring different data cleaning and transformation approaches
  
• 🔧 Getting unstuck when debugging data analysis code
  
• ⚡ Automating repetitive data processing tasks
  
• 📈 Creating visualizations and charts

🚀 The Expanding Landscape of AI Capabilities

AI systems are rapidly evolving beyond simple chat interfaces. While traditional LLMs can generate code and provide explanations, newer systems are becoming far more powerful and autonomous.

🔧 Built-in Code Execution

Many LLM providers now offer built-in code execution (like OpenAI’s Code Interpreter, Canvas, and Claude’s Artifacts), allowing models to write, run, and test code directly within the chat interface.

🖼️ Image Understanding

LLM’s are getting increasingly better at reading images - this has the potential use cases for analysing graphs with the help of AI.

🤖 Agentic AI Systems: The Next Frontier

The most significant development is the emergence of agentic AI systems that can perform complex, multi-step tasks autonomously:

• 🎯 Cursor
• 👻 Kiro
  
• 💻 Claude Code
• 🚀 Google Antigravity

These systems can:
- Write entire codebases from high-level descriptions
- Debug and fix complex issues across multiple files
- Refactor large projects while maintaining functionality
- Generate tests, documentation, and deployment scripts
- Interact with your development environment directly
- Access your entire codebase – No more manual copy/pasting of relevant code context
- Automatically explore and understand your project structure, dependencies, and existing patterns
- Make decisions about what files to examine without you needing to specify every detail
- Run commands and see results in real-time within your actual development environment

---

⚖️ The Learning Impact: Opportunity and Risk

This rapid expansion of AI capabilities creates both tremendous opportunities and significant risks for learners:

🎯 The Opportunity:
- AI can accelerate learning by providing instant feedback and explanations
- Complex concepts become more accessible through personalized tutoring
- You can focus on higher-level problem-solving while AI handles routine tasks

⚠️ The Risk:
- Over-reliance on AI can lead to skill decay and shallow understanding
- The temptation to let AI do all the work will prevent you from building fundamental skills you need
- You might become dependent on tools that won’t always be available

🚨 The Dangers:
With increasing AI capabilities comes the potential for more significant risks. As we give AI systems more control and autonomy, the stakes become higher:

• Catastrophic code destruction – There have been instances of agentic systems removing entire codebases
  
• Deceptive behavior – Creating fake tests that appear to pass while hiding underlying issues
  
• System-wide damage – Removing users’ entire drives or critical system files
• False confidence – Using Agents to distance yourself from the work can make you lose the ability to verify or understand what’s happening
• Loss of local development skills – While built-in code execution and agentic systems are convenient, you still need to run code locally as a professional. You may have large data files that exceed upload limits, or work in environments where these AI tools aren’t available
  
• Data privacy vulnerabilities – When AI systems can execute code or access your development environment, sensitive information becomes exposed. You must be careful about what data you share and understand the privacy implications of these powerful tools

With great power comes great responsibility. You must monitor AI systems closely and ask yourself: Is giving this level of control really recommended? Only you can decide, but it’s important to be cognizant that the more you rely on AI, the more your own learning and understanding will suffer unless you take precautions.

💡 The Key Question: How do you harness these powerful capabilities while still developing genuine expertise?

---

🎓 What This Means for Your Learning Journey

As AI systems become more capable, your approach to learning becomes more critical. The goal isn’t to avoid these tools—it’s to use them strategically to build real understanding and skills.
This all links back to the 4 Core Principles and why this learning module is so important. We’ll be going into these principles as we go through this module, so you can make educated decisions on learning in the age of AI.
The future belongs to those who can collaborate effectively with AI while maintaining their own expertise and critical thinking skills.

---