Why Does Bard Struggle with Context for Coding?

[bomb]

Bard provided high quality code when I described desired features. Its suggestions aligned with good design patterns. However, longer sessions exposed weaknesses - Bard struggled to maintain context and often misunderstood my intent. GPT-4 appears superior for extended coding help.

While promising, Bard needs better conversational context and comprehension to excel at collaborative programming. What's been your experience using Bard for coding tasks?

 

[Catalina]

I'm also trying to use the Bard API and Python to get JSON output for custom applications. However, the responses often don't match my prompts. GPT provides higher quality, on-topic JSON that's better for further processing.

Bard's API output needs improvement - it should follow prompt instructions more precisely. Have others found good techniques for getting usable JSON data from Bard for external applications? The API functionality feels limited compared to GPT right now.

 

[jitendra05]

Brad provides the coding syntax and parameters with accuracy like Chatgpt but it needs to optimize in the debugging process since it may misunderstand the real intentation sometimes as mentioned above in the above comment.