- The 800 developer survey reveals that more than half believe that AI can code better than humans
- But most are positive about the changes that AI is bringing
- Reservations remain, with job security, data privacy and inaccuracies that still cause concerns
With the great language models, they become more and more an inevitable characteristic in software spaces, and the new research of Clutch has revealed just over half (53%) of developers already think that LLM can better encode better than most humans.
As with many industries, artificial intelligence tools are widely adopted in the software world, with 49% of developers and team leaders who report that they use these articles every day, and 78% of developers use them several times a week.
Perhaps surprisingly, the study found an overwhelmingly favorable inclination when it comes to AI, with 42% positive about its use, compared to only 10% of respondents with worries and 8% who are “skeptical.”
Persistent fears
Although most developers hope to use AI, the study revealed some serious reserves. Mainly, developers are concerned about the management of AI confidential data, with 24% citing data privacy as their main concern.
However, not only that, since some (14%) care about the displacement of work, as well as inaccuracy (14%) and a loss of creativity (13%), while only 8%do not worry at all.
The labor market is rapidly evolving with the integration of AI, since 79% say that the skills of the soon will be a requirement for developers, and 45% believe that AI will reduce the barrier for Junior developers in the future.
More than two thirds (37%) think that the new encoders could be replaced or surpassed by AI, and after repeated news of layoffs of the technological sector, software engineers are increasingly concerned about the ruining the labor market.
The market for Junior developers seems increasingly difficult, with the entry level work that disappears thanks to LLMS, leaving a potentially disastrous scenario once the higher level developers retire in which companies remain with very little real experience, depending largely on AI without experts trained to supervise work.
Even now, more than half of the developers (59%) admit to trusting the code generated by the AI that they do not completely understand. That said, most do not see AI as a threat, with 80% that describes tools as facilitators, perhaps indicating a preference for short -term productivity on long -term planning.
