Urgent.News

What's breaking now, across thousands of outlets.

AI

Stack Overflow survey finds devs hooked on AI, but not totally sold on its judgment

Source attribution matters to 79% of respondents – handy for a company licensing its Q&A archive to model builders

Stack Overflow survey finds devs hooked on AI, but not totally sold on its judgment

Stack Overflow's latest survey reveals that developers are increasingly reliant on AI tools, but the technology remains controversial. Almost a third of developers surveyed spend four hours or more daily using AI, with 80 percent using the tools for at least an hour each day. Despite this, trust in AI output for critical work decisions is low, with only 7 percent of developers trusting AI for important work decisions.

Trust increases when the output can be easily verified, with 48 percent of developers trusting AI under these conditions. Source attribution plays a crucial role in determining trust, with 79 percent of developers considering it important. Stack Overflow CEO Prashanth Chandrasekar emphasized the skepticism among developers, stating that as AI agents become more autonomous, this skepticism acts as a safety net.

Developers need to know the origin of AI-generated answers, the context that informed their creation, and most importantly, the ability to verify their accuracy and relevance. The survey found that coding assistants, general-purpose chat tools, and automated agent workflows are the most commonly used AI technologies, with 66 percent, 63 percent, and 26 percent of respondents, respectively, employing these tools.

However, 17 percent of developers do not use AI tools at all. The commercial relationship between Stack Overflow and AI companies has been contentious. In early 2024, Stack Overflow introduced OverflowAPI, offering paid access to its question-and-answer content and requiring AI providers to attribute answers to relevant posts. This move led to user backlash, with some developers opting to remove their posts rather than have them used to train OpenAI models.

Despite the controversy, 52 percent of respondents reported learning a new coding skill or programming language this year, with AI code-generation tools being among the most common learning resources.

Written by urgent.news from The Register Science's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Also reported by 1 other outlet

Read the original at theregister.com →

More in AI

TouchGrass AI: Local Gemma 2B Micro-Adventure Generator

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass What I Built TouchGrass AI is a lightweight, offline-first web application that generates hyper-local, 3-step…

  • TouchGrass AI is an offline-first web app for generating hyper-local micro-adventures.
  • Users input environment and time to receive unique real-world observation quests.
  • Developed with Gemma 2B model locally via Ollama, and .NET 10 Minimal API backend.

More from Saturday 10 October →