A hot potato: Generative AI is disrupting a lot of areas, especially the programming industry. For a while now, tech leaders have claimed that the technology can make anyone a coder, and that kids don't need to learn these skills anymore. Some have even called English the hot new programming language.
Being able to code has required an understanding of languages such as Python, C++, or Java, notes Analytics India, but generative AI tools like GitHub Copilot and ChatGPT are making those skills less important.
Nvidia boss Jensen Huang is one of the most vocal proponents of using natural language prompts to code, but then he would be, considering he's the CEO of the company behind most of the world's advanced AI hardware. "Literally everyone can program a computer," he said at Berkeley Haas University's Dean's Speaker Series last year. "We have democratized computing in a very, very large way."
At the start of 2023, Andrej Karpathy, senior director of AI at Tesla, wrote in a post that the hottest new programming language is English.
The hottest new programming language is English
– Andrej Karpathy (@karpathy) January 24, 2023
Microsoft head Satya Nadella is another to extol the virtues of AI and AI-assisted programming. Like Huang, he calls this the democratization of coding. Microsoft owns GitHub Copilot through its GitHub subsidiary, of course. Then there's Sundar Pichai, who recently revealed that 25% of new code Google creates is written by AI.
Calling English the hot new programming language might be hyperbole, but there's no denying the effect AI is having on coding boot camps and active job postings for software developers, which have dropped 56%.
However, it might not be time to follow Huang's advice and drop programming in favor of more important skills such as biology, education, manufacturing, or farming.
"Coding" was never the source of value, and people shouldn't get overly attached to it. Problem solving is the core skill. The discipline and precision demanded by traditional programming will remain valuable transferable attributes, but they won't be a barrier to entry.
Many… https://t.co/0JmpGWfaT0
– John Carmack (@ID_AA_Carmack) February 26, 2024
While each new iteration of generative AI tools tends to be better at writing code than the last, some experts say AI tools still lack the essential coding skills that even junior programmers possess, and that LLMs like GPT-4 continue to make mistakes and are incapable of performing some tasks well, if at all.
Experts also say that a high level of coding is necessary when it comes to complex, large-scale software. Moreover, it would help if someone understood the code an AI generated so they could check it and correct any mistakes before the code is implemented.
It's also noted that some skill is needed to craft an accurate and precise prompt for an AI to generate what is required. As Huang put it, "There is an artistry to prompt engineering. It's how you fine-tune the instructions to get exactly what you want."
Some in the industry say AI is good for programming, helping speed up the process. About 60% of 65,000 developers surveyed in May by StackOverflow say they had used AI coding tools this year. The assistants may not have boosted their productivity or effectively helped prevent burnout, though, according to a recent study.