- Survey of 800 developers reveals over half believe AI can code better than humans
- But most are positive about changes AI is bringing
- Reservations remain, with job security, data privacy, and inaccuracies still causing concerns
With Large Language Models increasingly becoming an unavoidable feature in software spaces, and new research by Clutch has revealed just over half (53%) of developers already think LLMs can code better than most humans.
As with many industries, AI tools are widely adopted in the software world, with 49% of senior developers and team leads reporting they use such items every day, and 78% of developers use them several times a week.
Perhaps surprisingly, the study found an overwhelmingly favorable leaning when it comes to AI, with 42% feeling positive about its use – compared to just 10% of respondents with concerns, and 8% who are ‘sceptical’.
Lingering fears
Although most developers are looking forward to using AI, the study did reveal some serious reservations. Primarily, developers are worried about AI’s handling of sensitive data, with 24% citing data privacy as their top concern.
Not just that though, as some (14%) worry about job displacement, as well as inaccuracy (14%), and a loss of creativity (13%) – whilst only 8% are not concerned at all.
The job market is rapidly evolving with AI integration, as 79% say AI skills will soon be a requirement for developers, and 45% believe AI will lower the barrier for junior developers in future.
Over two thirds (37%) think new coders could be replaced or out-performed by AI – and after repeated news of tech-sector layoffs, software engineers are increasingly worried AI is ruining the job market.
The market for junior developers looks increasingly difficult, with entry-level jobs disappearing thanks to LLMs, leaving a potentially disastrous scenario once senior level developers retire in which companies are left with very little real experience – heavily relying on AI without trained experts to oversee the work.
Even now, over half of developers (59%) admit to relying on AI-generated code that they don’t fully understand. That being said, most don’t see AI as a threat, with 80% describing the tools as enablers – perhaps indicating a preference for short-term productivity over longer-term planning.
You might also like
- Take a look at our picks for the best firewall software around
- Check out our choice for best endpoint protection software to keep you safe
- 70% of people are sick of talking to AI – where did all the humans go?