AI is not ready to replace programmers in the short term


The developer community is wondering about the place of artificial intelligence in the coding process. A very extensive Stackoverflow study shows that almost three quarters of them use or will use AI tools. The majority of developers prefer to believe (39.3%) in the precision of the results, and 30.68% remain agnostic.

Since Open AI’s ChatGPT broke into the media field in November 2022, other players have followed suit. Co-pilot from Microsoft, Pitchfork or AI Developer Assistance from Google are sharpening their guns and constantly improving their coding assistance tools. Stackoverflow’s 2023 annual survey is informative as it covers a large sample of 90,000 developers, independent or working in companies of all sizes, with 40% of the panel working in an organization with fewer than 100 employees. Research mainly focuses on the use of artificial intelligence and machine learning (ML) as development support tools, as well as the coder’s perception of future development.

OpenAI launches its language model (LLM) with step-by-step coding expert development processes. Currently, the task of the help tool is mainly auto-completion.

According to a Stackoverflow survey, 70% of respondents are using or plan to use AI tools in their development process this year. It will come as no surprise to learn that beginners learning to code are more likely than professional developers to use AI tools
(82% vs. 70%).

To date, it doesn’t appear that coding professionals see a threat from job support functions as 77% of those surveyed are favorable or very favorable toward AI tools. Increased productivity is the main advantage, a criterion highlighted by 32.81% of developers.

More than 4 in 10 developers trust the accuracy of coding support tools

As for the precision of the results it gives AI tools, developers are divided. About 42% of them trust their accuracy, while 31% remain unsatisfied. Most coders prefer to trust them (39.3% machine), and 30.68% remain agnostic. Please note that 5.46% of respondents are highly suspicious of the accuracy of the results as shown in the chart below.

People who currently use AI tools mainly cite the benefits of writing code (82.5%), while debugging – correcting errors – and helping to write code represent only 48.9% of responses. This is followed by documenting the code (34.37%) or learning it (30%).

Developers estimated that their tasks will be different in a year – the survey was published in May 2023 – thanks to AI tools. Specifically, when it comes to writing code, 19% believe their work will be significantly different, and 52% believe it will change significantly.
For debugging and help, the shares increase to 31% (very different) and 46% (significant change).

According to the results of the Stackoverflow study, AI development assistance does not seem to worry coding experts too much, but given the rapid progress of ChatGPT and others, a significant transformation is to be expected in the coming years.



Source link

Leave a Comment