Interesting. I should have read the comments BEFORE adding mine. I am wondering if folks who program (I am one) have more tolerance for iteration, and hence are less troubled by initial "bugs" in LLM responses.
Interesting. I should have read the comments BEFORE adding mine. I am wondering if folks who program (I am one) have more tolerance for iteration, and hence are less troubled by initial "bugs" in LLM responses.
Again, I'm not a coder, but all coding is iterative, isn't? Write code, test, fix mistakes, write more code, repeat . . .
A YouTube Python expert named Corey Shafer made a video about using ChatGPT for coding last spring. It actually came out after I used ChatGPT to write my first Python scripts, so I felt pretty smug. But he provided a lot of suggestions for coders as to how they could make use of the tool. The thing he personally liked to do was, when facing a blank screen while starting a new project, ask ChatGPT to get him started, so he had something to work on.
In then end he determined while LLMs will not replace programmers, programmers who use LLMs will replace programmers who do not.
But my initial experiences with Acronic's Claude suggest he might have been too optimistic. I am pretty sure that within 18 months we will all regard ChatGPT 3.5 as a pretty crude and basic implementation of an LLM.
Interesting. I should have read the comments BEFORE adding mine. I am wondering if folks who program (I am one) have more tolerance for iteration, and hence are less troubled by initial "bugs" in LLM responses.
Again, I'm not a coder, but all coding is iterative, isn't? Write code, test, fix mistakes, write more code, repeat . . .
A YouTube Python expert named Corey Shafer made a video about using ChatGPT for coding last spring. It actually came out after I used ChatGPT to write my first Python scripts, so I felt pretty smug. But he provided a lot of suggestions for coders as to how they could make use of the tool. The thing he personally liked to do was, when facing a blank screen while starting a new project, ask ChatGPT to get him started, so he had something to work on.
In then end he determined while LLMs will not replace programmers, programmers who use LLMs will replace programmers who do not.
But my initial experiences with Acronic's Claude suggest he might have been too optimistic. I am pretty sure that within 18 months we will all regard ChatGPT 3.5 as a pretty crude and basic implementation of an LLM.