SE Radio 711: Scott Hanselman on AI-Assisted Development Tools

AI-assisted development tools and their impact on software engineering are explored, with Scott Hanselman, Vice President of Developer Community at Microsoft, providing historical context and practical advice. The conversation tackles the ambiguity introduced by AI in coding, contrasting it with the specificity of traditional programming. Hanselman argues that while AI can eliminate toil, it necessitates strong engineering judgment and a solid understanding of computer science fundamentals. Examples include using AI to modernize a 20-year-old blogging tool and the importance of test-driven development to verify AI-generated code. The discussion also covers the responsible use of AI agents, balancing productivity with security, and the potential of local models for specific coding tasks.

Outlines

Part 1: Introduction and Evolution

Part 2: AI as a Development Partner

Part 3: Verification, Testing, and Standards

Part 4: Workflow Optimization and Context

Part 5: Automation and Advanced Loops

Part 6: Security, Governance, and Language Choice

Part 7: Conclusion and Outlook

Sign in to continue reading, translating and more.

Continue

Software Engineering Radio - the podcast for professional software developers

Part 1: Introduction and Evolution

Introduction to AI-Assisted Development Tools and Overcoming Overwhelm

The Evolution of AI-Augmented Coding: From Syntax Highlighting to Agentic Loops

The Illusion of One-Shot Solutions and the Importance of Specificity in Programming

From Assembly to Prose: The Evolving Abstraction Layers in Programming

Part 2: AI as a Development Partner

The Value of Specificity and Brainstorming with AI: A Shipping Automation Example

The End of Toil: AI's Impact on Software Engineering and the Rise of Taste and Judgment

Balancing AI Assistance with Foundational Computer Science Knowledge

Part 3: Verification, Testing, and Standards

Verifying LLM Assertions: Fear-Driven Development and the Importance of Testing

The Responsibility of Shipping Code: Verification Loops and Security Considerations

Balancing Personal Projects with Professional Standards: The GeoCities Moment

From Vibe Coder to AI-Augmented Engineer: Architectural Considerations and Long-Term Maintainability

Part 4: Workflow Optimization and Context

Defining Success: The Importance of Clear Objectives and Verifiable Loops

Recognizing and Breaking Free from AI Loops: Defining Success and Providing Tools

Providing Context and Tools: Empowering AI with the Resources It Needs

Managing Context and Memory: Strategies for Long-Running AI Conversations

Context Window Management: Balancing AI and Human Understanding for Code Inheritance

Part 5: Automation and Advanced Loops

Ralph Loops: Automating Toil with Persistent and Stubborn Iteration

Building Your Own While Loop: Defining Stop Conditions and Mixing Models

The Economics of AI: Valuing Time and Finding Joy in Automation

Local Models: The State of Offline AI and the Potential for Specificity

Part 6: Security, Governance, and Language Choice

Security and Permissions: Balancing Power with Responsibility in AI Agents

Managing AI Agents: Balancing Convenience with Security and Control

Security in Practice: Sandboxing, Auditing, and the Spectrum of Risk

Strongly Typed Languages: Enhancing Correctness and Reducing Fear

Language Choice: Balancing Portability, Delight, and the Right Tool for the Job

Part 7: Conclusion and Outlook

The Craft Remains: AI's Potential to Eliminate Toil and Enhance Creativity

Podcast Recommendations and Closing Remarks

SE Radio 711: Scott Hanselman on AI-Assisted Development Tools

Software Engineering Radio - the podcast for professional software developers

Part 1: Introduction and Evolution

00:00Introduction to AI-Assisted Development Tools and Overcoming Overwhelm

Introduction to AI-Assisted Development Tools and Overcoming Overwhelm

00:53The Evolution of AI-Augmented Coding: From Syntax Highlighting to Agentic Loops

The Evolution of AI-Augmented Coding: From Syntax Highlighting to Agentic Loops

04:10The Illusion of One-Shot Solutions and the Importance of Specificity in Programming

The Illusion of One-Shot Solutions and the Importance of Specificity in Programming

07:30From Assembly to Prose: The Evolving Abstraction Layers in Programming

From Assembly to Prose: The Evolving Abstraction Layers in Programming

Part 2: AI as a Development Partner

11:07The Value of Specificity and Brainstorming with AI: A Shipping Automation Example

The Value of Specificity and Brainstorming with AI: A Shipping Automation Example

13:32The End of Toil: AI's Impact on Software Engineering and the Rise of Taste and Judgment

The End of Toil: AI's Impact on Software Engineering and the Rise of Taste and Judgment

15:13Balancing AI Assistance with Foundational Computer Science Knowledge

Balancing AI Assistance with Foundational Computer Science Knowledge

Part 3: Verification, Testing, and Standards

18:05Verifying LLM Assertions: Fear-Driven Development and the Importance of Testing

Verifying LLM Assertions: Fear-Driven Development and the Importance of Testing

21:39The Responsibility of Shipping Code: Verification Loops and Security Considerations

The Responsibility of Shipping Code: Verification Loops and Security Considerations

24:01Balancing Personal Projects with Professional Standards: The GeoCities Moment

Balancing Personal Projects with Professional Standards: The GeoCities Moment

25:29From Vibe Coder to AI-Augmented Engineer: Architectural Considerations and Long-Term Maintainability

From Vibe Coder to AI-Augmented Engineer: Architectural Considerations and Long-Term Maintainability

Part 4: Workflow Optimization and Context

29:32Defining Success: The Importance of Clear Objectives and Verifiable Loops

Defining Success: The Importance of Clear Objectives and Verifiable Loops

32:33Recognizing and Breaking Free from AI Loops: Defining Success and Providing Tools

Recognizing and Breaking Free from AI Loops: Defining Success and Providing Tools

34:49Providing Context and Tools: Empowering AI with the Resources It Needs

Providing Context and Tools: Empowering AI with the Resources It Needs

35:58Managing Context and Memory: Strategies for Long-Running AI Conversations

Managing Context and Memory: Strategies for Long-Running AI Conversations

38:08Context Window Management: Balancing AI and Human Understanding for Code Inheritance

Context Window Management: Balancing AI and Human Understanding for Code Inheritance

Part 5: Automation and Advanced Loops

40:37Ralph Loops: Automating Toil with Persistent and Stubborn Iteration

Ralph Loops: Automating Toil with Persistent and Stubborn Iteration

43:40Building Your Own While Loop: Defining Stop Conditions and Mixing Models

Building Your Own While Loop: Defining Stop Conditions and Mixing Models

45:23The Economics of AI: Valuing Time and Finding Joy in Automation

The Economics of AI: Valuing Time and Finding Joy in Automation

48:22Local Models: The State of Offline AI and the Potential for Specificity

Local Models: The State of Offline AI and the Potential for Specificity

Part 6: Security, Governance, and Language Choice

50:07Security and Permissions: Balancing Power with Responsibility in AI Agents

Security and Permissions: Balancing Power with Responsibility in AI Agents

52:38Managing AI Agents: Balancing Convenience with Security and Control

Managing AI Agents: Balancing Convenience with Security and Control

55:31Security in Practice: Sandboxing, Auditing, and the Spectrum of Risk

Security in Practice: Sandboxing, Auditing, and the Spectrum of Risk

57:09Strongly Typed Languages: Enhancing Correctness and Reducing Fear

Strongly Typed Languages: Enhancing Correctness and Reducing Fear

58:12Language Choice: Balancing Portability, Delight, and the Right Tool for the Job

Language Choice: Balancing Portability, Delight, and the Right Tool for the Job

Part 7: Conclusion and Outlook

1:00:08The Craft Remains: AI's Potential to Eliminate Toil and Enhance Creativity

The Craft Remains: AI's Potential to Eliminate Toil and Enhance Creativity

1:01:27Podcast Recommendations and Closing Remarks

Podcast Recommendations and Closing Remarks