The fact that artificial intelligence can bring results, from extremely impressive to shockingly problematic, can explain why developers seem so divided on this technology. Wired surveyed programmers in March, to ask how they feel about AI coding, and said that the enthusiastic proportion regarding AI tools (36 percent) was reflected by the part that felt skeptical (38 percent).
“Undoubtedly AI will change the way the code is produced,” says Daniel Jackson, IT specialist, who is currently investigating how to integrate artificial intelligence with large-scale software development. “But it would not surprise me if we were disappointed” that the noise would pass.
Jackson warns that AI models are fundamentally different from compilers that transform the code written in a high level into a lower level language, which is more productive in using machines, because they do not always follow the instructions. Sometimes the AI model can take instructions and perform better than a developer – at other times it can be much worse.
Jackson adds that the atmosphere coding falls when someone builds sedate software. “There are almost no applications in which” mainly works “is good enough,” he says. “As you will soon take care of the software, you care that it works well” –
Many software projects are complicated, and changes in one code section can cause problems elsewhere in the system. Jackson says that experienced programmers are good in understanding a wider picture, but “wonderful language models cannot justify their type of dependence.”
Jackson believes that software creation can evolve with more modular code databases and a smaller number of dependencies to the AI Blind Glots room. He expects AI to replace some programmers, but he will also force much more to think about their approach and focus more on project design.
Too much relying on artificial intelligence can be “a little approaching disaster” – adds Jackson, because “we will not have mass code, full of security, but we will have a new generation of programmers that will not be able to cope with these threats.”
Learn to cod
Even companies that have already integrated coding tools with the software creation process say that the technology remains too unbelievable to be wider.
Christine Yen, general director of Honeycomb, a company that provides technology to monitor the performance of gigantic software systems, claims that projects that are uncomplicated or formal, such as component libraries, are more susceptible to the employ of AI. Despite this, she says that programmers from her company, who employ artificial intelligence in their work, increased their efficiency by about 50 percent.
Yen adds that in the case of everything that requires a good judgment, in which performance is crucial, or when the resulting code affects sensitive systems or data: “To be honest, it is not good enough to be additive.”
“A difficult part of building software systems not only I write a lot of code,” he says. “Engineers will still be necessary, at least today, to have this treatment, judgment, tips and direction” –
Others suggest that there is a change in labor. “We don’t see less demand for programmers,” says Liad Elidan, general director of Milestone, a company that helps companies measure the impact of AI generative projects. “We see a lower demand for average or low programmers.”
“If I build a product, I could need 50 engineers, and now maybe I only need 20 or 30” – says Naveen Rao, vice president of AI at Databicks, a company that helps gigantic companies build their own AI systems. “This is absolutely true.”
Rao says, however, that learning coding should remain a valuable skill for some time. “It’s like saying” not teaching your kid to learn mathematics, “he says. He adds that understanding how to best use computers can remain extremely valuable.
Yegge and Kim, experienced coders, believe that most programmers can adapt to the upcoming wave. In their book on VIBE coding, the pair recommends new software creation strategies, including modular code bases, continuous tests and lots of experiments. Yegge claims that the use of artificial intelligence for writing software is evolving in its own – somewhat risky form of art. “It’s about how to do it without destroying the difficult disk and exhaustion of the bank account,” he says.