WeAreDevelopers LIVE - the weekly developer show with Chris Heilmann and Daniel Cranney

Is AI creating a hidden productivity tax for developers? We explore why trust in AI tools is decreasing as usage rises.

#1about 2 minutes

Tips for delivering conference talks for online recordings

Speakers should avoid in-person-only references like pointing to booths or asking for a show of hands to ensure the talk remains relevant for online viewers.

#2about 5 minutes

Understanding the Global Black Belt role at Microsoft

The Global Black Belt team acts as subject matter experts on developer productivity and DevOps, engaging in strategic customer projects and exploring edge cases.

#3about 4 minutes

Why tech stack migrations are often driven by hype

Many language migration decisions are influenced more by industry trends than proven outcomes, which can lead to costly rollbacks if the new technology lacks ecosystem support.

#4about 10 minutes

The hidden productivity tax of AI-generated code

AI coding assistants can introduce a productivity tax because developers spend significant time debugging almost-correct code and correcting AI hallucinations.

#5about 4 minutes

Lessons from the Sketch outage caused by LLM code

The Sketch outage highlights the risks of deploying LLM-generated code without rigorous, deterministic security checks, as AI is trained on human code that contains vulnerabilities.

#6about 7 minutes

Why developer trust in AI coding tools is declining

As AI tool usage increases, developer trust is falling due to the overhead of debugging, the pressure to use mandated tools, and the negative impact on junior developer learning.

#7about 5 minutes

Why jQuery remains a popular web framework

Despite its age, jQuery's persistence is due to its integration in legacy systems like WordPress and its original promise of simplifying complex tasks, similar to the current AI hype.

#8about 6 minutes

Customizing developer environments and managing side projects

A discussion on personalizing code editors with themes and fonts, and the common developer habit of accumulating domains for side projects that often get abandoned.

#9about 15 minutes

Using a multi-agent AI framework to migrate COBOL

Migrating legacy COBOL code is challenging due to a lack of public training data and hardware dependencies, requiring a multi-agent AI framework for reverse engineering and transformation.