Clint Hocking, the former director of Assassin’s Creed Hexe and Far Cry 2, recently revealed that AI coding tutors failed him badly when he tried to learn JavaScript with ChatGPT’s help. In a feature published in Edge Magazine about AI’s role in games development, Hocking admitted to using ChatGPT as a learning tool—and regretted nearly every minute of it. His blunt assessment: “It was brutal. ChatGPT kind of sucked. It didn’t really know how to code. Everything was broken”.
Key Takeaways
- Clint Hocking used ChatGPT to learn coding and found the experience deeply frustrating and inefficient.
- ChatGPT generated broken code that Hocking had to debug without knowing how to code himself.
- After roughly six months of late-night debugging sessions, Hocking eventually learned JavaScript despite ChatGPT, not because of it.
- Hocking no longer uses AI tools for coding work.
- He believes AI integration in games production is inevitable, even though his personal experience was negative.
Why AI coding tutors failed Hocking
The core problem was simple: ChatGPT generated code that didn’t work, and Hocking lacked the foundational knowledge to fix it. “It was mostly me trying to debug code without knowing how to code myself,” he explained. This is the inverse of how learning should work. A tutor should clarify concepts and guide problem-solving. Instead, Hocking became trapped in a loop of trial-and-error, attempting to reverse-engineer what ChatGPT had written without understanding the underlying logic. The AI tool became an obstacle rather than a shortcut.
What makes this experience particularly revealing is that Hocking is not a beginner to complex systems. He has shipped AAA games and directed teams of hundreds. Yet he still found AI coding tutors inadequate for basic skill acquisition. If an experienced creative director struggles this badly, what does that say about ChatGPT’s utility for actual novices trying to break into programming?
The brutal six-month debugging marathon
Hocking’s path to learning JavaScript was unconventional and punishing. He spent roughly half a year untangling ChatGPT’s mistakes, often working late into the night. That is not a learning curve—that is a slog. Traditional coding education, even self-taught through documentation and Stack Overflow, typically moves faster because humans writing tutorials understand pedagogy. They anticipate common mistakes and explain why code works. ChatGPT does neither. It generates plausible-looking code that fails in subtle ways, leaving learners to puzzle out the difference between syntax errors, logic errors, and architectural misunderstandings.
The silver lining: Hocking eventually did learn JavaScript. But the route was inefficient enough that he abandoned AI coding tutors entirely. He no longer uses them for coding work. That decision speaks louder than any benchmark score. When a developer with his background and patience decides a tool is not worth the friction, it is a signal that AI coding tutors are not yet ready for their intended use case.
AI in games production remains inevitable—for now
Despite his poor experience, Hocking told Edge Magazine that he believes AI integration in games development is “inevitable”. This is not a prediction of quality or desirability—it is a statement about industry momentum. Studios are experimenting with AI for asset generation, animation, dialogue, and code scaffolding. Some experiments will fail. Others will stick. The question is not whether AI will be used, but how responsibly it will be deployed and whether developers will be honest about its limitations.
Hocking’s experience is a useful counterweight to hype. AI coding tutors are not yet reliable enough to replace human instruction, documentation, or pair programming with experienced developers. They can accelerate certain workflows for people who already know how to code—spotting bugs, suggesting refactors, generating boilerplate. But as a learning tool for beginners, they remain frustratingly broken. The industry should take note.
Is ChatGPT a good tool for learning to code?
Not yet, according to Hocking’s experience. ChatGPT generates code that looks correct but often fails in ways that require debugging knowledge to understand. For someone without that foundation, the tool becomes a source of confusion rather than clarity. A human tutor, coding bootcamp, or structured online course is more efficient for beginners.
Will AI coding tutors improve?
Almost certainly. ChatGPT’s code generation has already improved since Hocking’s experience, and newer models may perform better. However, the core challenge remains: generating correct, pedagogically sound explanations is harder than generating working code. Until AI can explain why code works and anticipate learner misconceptions, it will remain inferior to human instruction.
Is Clint Hocking still involved in game development?
Hocking directed Assassin’s Creed Hexe and Far Cry 2, but the available reporting does not specify his current role or projects. His comments about AI in games development suggest ongoing engagement with industry trends, but no active title is confirmed in the available sources.
Clint Hocking’s honest assessment of AI coding tutors cuts through the noise. ChatGPT did not teach him to code—six months of painful debugging and late nights did. If you are considering using AI to learn programming, his experience suggests you would be better served by traditional resources: books, courses, mentors, and documentation written by humans who understand how to teach. AI will eventually earn a place in the developer’s toolkit, but it is not there yet.
Edited by the All Things Geek team.
Source: TechRadar
