Software developers and engineers are experimenting with generative A.I. tools such as ChatGPT and Google’s Bard. But are they actually integrating these tools into their workflows?
According to Stack Overflow’s latest Developer Survey, which queried 89,184 developers from 185 countries, the answer to that question is “yes.” Among developers using A.I. tools, writing code seems to be the most popular pursuit, followed by debugging and documentation. Check out the full chart:
However, not all developers trust these tools, with 27.17 percent saying they either somewhat or highly distrust the accuracy of A.I. tools’ output, compared to 30.68 percent who neither trust nor mistrust, and 42.15 percent who either somewhat or highly trust.
In general, developers feel that A.I. tools could speed up their workflows and learning, boost efficiency, and improve their coding accuracy. But it’s still early days for many A.I.-platforms; while chatbots can write and debug code, for example, tech pros should still closely examine the output for accuracy.
The other big question is whether A.I. will eliminate a hefty percentage of developer and engineer jobs. According to a recent report by Goldman Sachs, some 29 percent of computer and mathematical jobs are potentially vulnerable to A.I. takeover. Even so, many tech pros also think that automation will free up their time to focus on more interesting and creative pursuits, such as ideating new features.
During a recent episode of ‘Tech Connects,’ Nick Durkin, field CTO of harness.io (where he’s responsible for the organization's worldwide field engineering team, post-sales engineering team, and a portion of product), broke down why you shouldn’t necessarily fear A.I.:
As the A.I. market matures, we could see the rise of “app stores” where developers can sell A.I.-powered models and services. That could expand the reach of A.I. to new niches and subindustries. In the meantime, it’s probably worth experimenting with how chatbots and LLMs can impact your current workflow; just double-check the output before committing it to your current project.
