What Do Developers Want From AI? (Filed under DevEx)

What Do Developers Want From AI? (Rubric #DevEx)

Summer. 2024 It was an interesting year. article The evolution of AI is a turning point, but this is not the first time that humanity has faced technological revolutions that are changing the way people work. Therefore, the authors of the article decide to draw parallels with the development of the automotive industry and focus on the needs and goals of our developers. This is an article from the "Hunam centric approach to developer productivity" series from the guys at Google. beforehand. Now for the highlights of the article

1) Recent advances in AI have led to a large number of development tools to support artificial intelligence. (for example, DuetAI, CoPilot ChatGPT for programming tasks). After that, many studies were conducted on the impact of these tools on the productivity of developers, where questions were asked like this. Do AI enhancements increase the speed of code writing? Do they improve the quality of the written code? Do they help developers find more creative solutions? 2) But there has been much less discussion about how developers want to interact with AI in their tools. The authors of the study, with their focus on a human-centric approach to understanding developer productivity, decided to use this and conduct research to answer the question, where do developers want AI in their workflows, and what do they think its effects will be? 3) Here, the authors begin to draw parallels with cars to understand needs. Understanding what people want and need from technology is the foundation of user experience research, they say. However, this approach is sometimes questioned, especially when it comes to major technical innovations – in these cases, recall the famous phrase attributed to Henry Ford: “If I asked people what they wanted, they would say faster horses.” 4) To investigate the preferences of developers, the authors conducted numerous interviews and user follow-ups with developers with different experiences with AI-enabled development tools. Developers have expressed their desire for such tools to save their time and energy by helping them do their jobs more efficiently, meaning they want AI to support simpler tasks and reduce routine, allowing them to focus on solving complex problems and the creative aspects of their work. 5) In order to understand what areas prevent developers from being productive at Google, researchers conduct surveys and analyze logs. This allows you to focus your efforts on the most important factors, the three of which at Google right now.

• Technical debt Poor or missing documentation Difficulties in learning new platforms and technologies 6) The authors classified the evolution of AI improvements into three types (drawing parallels with the automotive industry) Improving existing human capabilities. For cars, it's power steering, anti-lock braking. For AI in development, this is code completion. - Human empowerment (addition). For cars, it's a rear-view camera, blind spot control. Code review suggestions, chatbots Delegation of human capabilities (Features that remove human in the loop). For cars, this is lane retention, automatic braking, improved cruise control. For AI in development it is automatic rollback of problematic releases, automatic deletion of dead code, generation of tests. 7) Successful implementation of changes with AI is necessary Understanding developer resistance and skepticism Creating the right infrastructure for implementation Ensuring safe and equitable access in all sectors Maintaining a focus on developer goals, not just technological capabilities

#Management #Leadership #Software #SoftwareDevelopment #Metrics #Devops #Processes #AI #ML #DevEx