ChatGPT (LLMs) replacing programmers? Ridiculous
- Thread starter Aquarians
- Start date
I find stack overflow piecemeal and time consuming to sift through.
I'm actually a decent programmer in C and Fortran. But to my old brain, Python feels like an ad hoc collection of arbitrary constructs thrown together.
Python is a horrible language, if you are coming from C try C#, you can do anything you want to with C# and it is way faster than Python, with the .net framework ready for web and desktop development.
Good answer!
Prompt: "ChatGPT (LLMs) replacing programmers? Ridiculous"
GPT 3.5 Answer:
The idea of large language models (LLMs) like ChatGPT replacing programmers entirely is indeed a far-fetched notion and could be considered ridiculous for several reasons:
Prompt: "ChatGPT (LLMs) replacing programmers? Ridiculous"
GPT 3.5 Answer:
The idea of large language models (LLMs) like ChatGPT replacing programmers entirely is indeed a far-fetched notion and could be considered ridiculous for several reasons:
- Complexity of Programming Tasks: While LLMs are powerful in generating human-like text based on prompts, programming involves much more than just writing code. It requires problem-solving, algorithmic thinking, understanding of data structures, debugging skills, and domain-specific knowledge, which are not yet within the capabilities of LLMs.
- Understanding Context and Intent: Programming often involves understanding the context of a problem and the intent behind a solution. LLMs may generate code snippets based on given prompts, but they lack the ability to fully comprehend the nuances of a programming task, including user requirements, system constraints, and potential edge cases.
- Quality and Reliability: The code generated by LLMs may not always be of high quality or reliable. Programming requires writing efficient, well-structured, and maintainable code, which may not always be achieved solely through text generation by LLMs.
- Creativity and Innovation: Programming often requires creativity and innovation to come up with novel solutions to complex problems. While LLMs can assist in generating code based on existing patterns, they may not be capable of true innovation or thinking outside the box.
- Human Expertise and Oversight: Even if LLMs were to assist in programming tasks, human expertise and oversight would still be essential. Developers bring a wealth of experience, domain knowledge, and critical thinking to the table, which cannot be replaced by AI systems.
Last edited:
LOL, good one
Unless Skynet spontaneously emerges from the combination of Nvidia chipsets, nothing of substance will change. Right now AI is as much conscious, intelligent and capable of autonomous operation as the Mechanical Turk was in the 19th century. Fascinating technology for the day but useless without a real human to back it.
What the vast majority don’t understand is that the LLM “hallucinations”, going off the reservation in response to prompts, seemingly elementary errors in output etc. are not small quirks which will get worked out of the model as it “learns”, but rather totally unavoidable results stemming from the system’s basic design.
Neither more compute nor more training data is capable of eliminating these problems - the latter especially, since more and more of the ingested content is itself botshit. Therefore, LLMs will remain unsuitable for all but a handful of tasks, and end up being a huge failure from an economic/business POV.
Neither more compute nor more training data is capable of eliminating these problems - the latter especially, since more and more of the ingested content is itself botshit. Therefore, LLMs will remain unsuitable for all but a handful of tasks, and end up being a huge failure from an economic/business POV.