On September 9, a new investigation conducted an in-depth analysis of Gene-based artificial intelligenceAIYeahsoftware developerTHE TRANSFORMATIVE ROLE OF DAILY WORK. THE RESULTS SHOW THAT THERE IS A CLEAR DIVISION AMONG ENGINEERS WITH DIFFERENT QUALIFICATIONS: SENIOR ENGINEERS PREFER TO USE AI IN LARGE NUMBERSCode, TAKES THE INITIATIVE TO AMEND THE CODE, AND CONSIDERS AI TO SAVE TIME; BY CONTRAST, PRIMARY DEVELOPERS ARE MORE CAUTIOUS AND ACCEPT THE AI TOOL AT A SLOWER PACE, WHICH ALSO MEANS THAT LESS EFFICIENCY GAINS ARE MADE。
The survey carried out by the Cloud Platform Fastly showed that a growing number of senior developers would not only generate a large number of codes using the AI tool, but would also deploy to a much higher proportion of the production environment than less qualified engineers。
THE SURVEY COVERED 791 PROFESSIONAL DEVELOPERS, WITH NEARLY ONE THIRD OF THE SENIOR ENGINEERS 10 YEARS OR MORE IN THE INDUSTRY INDICATING THAT MORE THAN HALF OF THE CODES THEY DELIVERED WERE GENERATED BY AI, WHILE THE PROPORTION OF PRIMARY DEVELOPERS 2 YEARS OR LESS WAS ONLY 13%, LESS THAN HALF OF THE SENIOR ENGINEERS。
THESE DATA NOT ONLY INDICATE THAT SENIOR ENGINEERS USE AI MORE FREQUENTLY, BUT ALSO REFLECT THEIR GREATER CONFIDENCE IN AI GENERATION CODE INPUT PRODUCTION。
AI NOTED THAT THE SURVEY ALSO REVEALED A GAP BETWEEN “SENSITIZATION SPEED” AND “REAL MODIFICATIONS”. THE DEVELOPERS OF 28% INDICATED THAT THEY OFTEN SPEND A GREAT DEAL OF TIME REPAIRING OR REWRITING THE CODES GENERATED BY AI, RESULTING IN AI BEING ALMOST OFFSET BY THE ADVANTAGES THAT IT WOULD HAVE BROUGHT; AND THE DEVELOPERS OF 14% STATED THAT THEY RARELY NEEDED TO MAKE SIGNIFICANT CHANGES TO THE AI CODES。
Nevertheless, over half of the respondents indicated that the AI tools, including GitHub Copilot, Google Gemini and Anthropic Claude, helped them to move up the pace of their work. Senior engineers were more motivated, and 59% found AI to have accelerated the pace of work, while the primary developers in this view were only 49%。
IT IS WORTH NOTING THAT EVEN THOUGH SENIOR ENGINEERS ADMITTED THAT MORE EFFORT WAS NEEDED TO CORRECT AI ERRORS, AMONG THEM, THE PROPORTION OF AI THAT COULD SAVE SIGNIFICANT TIME WAS STILL TWICE AS MUCH AS THAT OF THE PRIMARY DEVELOPERS。
THE DIFFERENCE BETWEEN SENIOR AND PRIMARY DEVELOPERS MAY NOT BE DUE TO THE LEVEL OF ENTHUSIASM FOR AI, BUT RATHER TO A GAP IN PROFESSIONAL COMPETENCIES. MORE THAN HALF OF THE PRIMARY DEVELOPERS BELIEVE THAT AI SUPPORT ONLY ALLOWS THEMSELVES TO “SLIGHT UP”, WHILE THE SENIOR ENGINEERS WHO SHARE THE SAME VIEW ARE ONLY 39%; INSTEAD, THE SENIOR DEVELOPERS OF 25% SAY AI LETS THEMSELVES “SLOW UP”, WHICH IS ABOUT TWICE AS HIGH AS THE PRIMARY DEVELOPERS。
Fastly offers a rational explanation: senior developers are better at discovering nuances in codes. With a lot of experience, they can identify the situation where the AI generation code is “face-right but actually operating abnormally”, which makes it more efficient to correct errors and avoid disruption of the work rhythm。
The survey also highlighted a contradiction in the prevalence of AI tools: Many developers think AI makes itself feel faster, but external research has come to the opposite conclusion. Before the results of Fastly ' s survey were released, there was a random comparison test earlier this summer that showed that when a senior open source developer used the code support tool, the time to complete the task was increased by 19%. According to Fastly, this difference may be due to a psychological factor - the rapid automatic completion of AI would create a “successful” illusion at an early stage, but follow-up would require considerable modifications that would eventually offset some of the forward advantage。
ALTHOUGH AI HAS PRODUCED UNEVEN EFFICIENCY GAINS, ITS POSITIVE IMPACT ON JOB SATISFACTION IS CLEAR. REGARDLESS OF THE LEVEL OF SENIORITY, THE DEVELOPER OF ABOUT 80% INDICATED THAT IT WOULD FEEL MORE INTERESTING TO WORK USING AI TO WRITE CODE。
FOR SOME DEVELOPERS, AI IS ATTRACTIVE TO REDUCE DUPLICATION OF WORK; FOR OTHERS, IT IS A NEW SENSE OF “REQUIRED GENERATION OF AVAILABLE CODES”. IN THIS INDUSTRY, WHICH IS PLAGUED BY PROFESSIONAL FATIGUE AND A BACKLOG OF TASKS, THERE MAY BE SIGNIFICANT VALUE IN THIS MOTIVATIONAL BOOST EVEN IF PRODUCTIVITY INCREASES ARE NOT CLEAR。
SUSTAINABILITY IS ANOTHER CENTRAL ISSUE REVEALED BY THE SURVEY. DATA SHOW THAT DEVELOPERS ARE INCREASINGLY CONCERNED ABOUT THE ENVIRONMENTAL COSTS OF AI, INCLUDING ITS LARGE CARBON FOOTPRINT. TWO THIRDS OF RESPONDENTS ACKNOWLEDGED THAT AI HAD A HIGH ENERGY DEMAND AND MOST INDICATED THAT GREEN PROGRAMMING PRACTICES HAD BEEN USED IN THEIR WORK. THIS PRACTICE HAS BEEN FOLLOWED BY AN INCREASE IN THE NUMBER OF PERSONS WITH HIGHER QUALIFICATIONS: THIS PROPORTION IS JUST IN THE MAJORITY OF PRIMARY DEVELOPERS, WHILE MID- AND SENIOR-LEVEL ENGINEERS ARE CLOSE TO 80%。