‘AI stalls at 80%’: Why AI isn’t coming for your software development job just yet, but is changing work

Reddit viral post provokes a conversation in the developer community about whether an artificial intelligence code assistants such as Github Copilot will finally replace software engineers – or simply redefine their roles.

The original post, the subscriber of technology -focused subreddit, is similar to AI’s help to the decades -long -time conflict to build self -driving cars. “I was historically from the real filth on this front, but more and more I think the AI ​​blog will become self-driving cars because it will get 95 % of the road there and then stumbled at 95 % for 15 years-and this is really 5 % important.”

The user expects a shift in the role of developers: “Our jobs will turn into a review of small parts of the code written from artificial intelligence all day and fix it if necessary.” While this may reduce the need for engineers in some organizations, they have argued that it could simultaneously create new roles because of “a group of non -technical people [trying] To write programs with artificial intelligence that will be the carts of animals. “

He repeated one of the highest classification of a self -identified software engineer, which confirms the benefit of tools while emphasizing their restrictions. They wrote: “Amnesty International/Copilot can do 80 % of the tasks at work. The last 20 % is much more difficult until it is done.” “Even with rewriting my 3-5X claims, I’m forced to take over.”

The commentator argued that the dream of replacing engineers in bulk is still far away-especially since artificial intelligence will need to explain mysterious or complex instructions from non-engineers. “Obtaining Amnesty International to replace software engineers will require this to perform the most difficult 20 % of tasks, correctly and as shown by people who are not software engineers.

Instead, artificial intelligence tools have become “complications of power”, as they said – which makes experienced developers faster and more productive. “When I face a specialized problem, I used to comb it with Stackoveflow 10-20 … Now with artificial intelligence, I can identify problems much faster.”

They also parallel to another digital revolution: “Excel was a loudspeaker for the accountants, and it was not replaced by accountants. We have more accounting jobs there now more than any time in the past.”

Perhaps the most important of them, they pointed to the accumulation of tremendous technical debts through institutional programs. “The major companies have trillions of men who deserve technical debt. If their employees become more efficient, a better service will be provided to use these employees more efficient than getting rid of them.”

The conversation reflects an increasing consensus in the world of technology: While artificial intelligence quickly converts the way the programs are written, reviewed and corrected, it is unlikely to replace full engineers behind the keyboard – at least any time soon. Instead, it represents a new age of augmented development, as speed skills and problem solving may be more important than ever.