With the news turning on multimillionaire agreements for new artificial intelligence companies and salary of workers of the multimillionaires, it was a study of a small non -profit research group that caught the attention in the technological world last week.
His findings were simple but surprising: AI made software engineers slow.
“When developers can use AI tools, they take 19% more to complete the problems, a significant slowdown that goes against the beliefs of developers and expert forecasts,” the non -profit group said in his report, Metr, which specializes in the evaluation of AI models.
“This gap between perception and reality is surprising: developers expected AI to accelerate it in 24%, and even after experiencing the deceleration, they still believed that AI had accelerated them by 20%,” added the authors of Metr.
The results can simply reflect the limits of current technology, they said, but they still offer a verification of reality about what could be said that it is the strongest part of the broad euphoric career: coding.
In the last year, new companies focused on generating software code have been the subject of an intense bidding war that has only intensified in recent weeks.
On Monday, the Windsurf coding company was acquired by another AI startup, cognition, after an agreement with Openai fell. Google screwed the Windsurf CEO while signing a license agreement of $ 2.4 billion. Cursor, which also focuses on the generation of AI code, was valued at $ 10 billion in a May financing round that generated $ 900 million. Vibos coding, a coding style that depends completely on AI, has already become part of the technological lexicon, and discussions about the future of developer work in most online were dedicated to technology can be found.
The talent of AI also remains in high demand, with the goal of the Facebook parents that offers multimillionaire payment days. LinkedIn discovered that “AI engineer” is the fastest growing work title among recent university graduates, with two related roles, data centers and systems engineer, in numbers 3 and 4.
The AI Gold Rush has arrived when the general work openings for software developers reached a minimum of five years earlier this year, asking questions about the responsibility of AI for the deceleration.
Among the most prominent companies that announce large rounds of layoffs have been Microsoft, whose CEO, Satya Nadella, has declared that up to 30% of the Microsoft code is now written by AI. Bloomberg News discovered that in a recent round of layoffs that occurred in the native state of Microsoft in Washington, software engineering was, with much, the largest individual work category to receive pink slippers, which represents more than 40% of the approximately 2,000 cut positions.
While it is clear that AI can write code, it is much less safe if technology raises a direct threat to encode short -term works.
In an article published on Wednesday, MIT researchers presented the “many” challenges that still exist before AI can really start replacing software engineers wholesale.
The main obstacles come when the AI programs are asked to develop a scale code, or with a more complex logic, the authors found.
“Everyone is talking about how we no longer need programmers, and now there is all this automation available,” said Armando Solar -lezama, Professor of the MIT and main author of the study, in a press release. “On the one hand, the field has progressed greatly. We have tools that are much more powerful than anyone we have seen before. But there is also a long way to go to get the complete promise of the automation we would expect.”
What problems exists in the current labor market of encoders may have more to do with the broader economic slowdown than with abrupt technological changes, experts say.
“The teams are becoming smaller,” said Heather Doshay, a member of Signalfire, a venture capital firm that invests in AI companies. “Not necessarily because of the AI, but for the demands of the market and the operating expenses. What is happening is that companies are asking: ‘How can we remain thin and hire less people while our financial track extend financially?'”
For a limited AI, many encoders remain anxious. A popular website that tracks technological dismissals shows that the pace of separations has increased in the last three quarters after seeing constant decreases over the previous six, although they remain well below a 2023 peak. In blind, an anonymous application of popular messages among technological workers, the issue of AI taking coding work is hot, with much skepticism about whether it is really happening, or if the narration is an excuse that has allowed companies to reduce staff.
Gareth Patterson, a 25 -year -old New York city resident, says he could transition from a sales role to engineering only after going through an exhausting and non -stop regime that had the temporary cost of most of his social life, not to mention his training schedule.
He says that the reward has been worth it because his salary now allows him to have available income in one of the most expensive cities in the world.
But do not envy those who try to break or even adapt to the new era.
“The expectations for an engineer are very high,” said a senior software engineer from a tax firm and audit. “Now we are only seeing that the upper talent is hired. It is intimidating.”
