In a recent study by Model Evaluation and Threat Research, experienced software developers anticipated that utilizing AI tools would enhance productivity. Instead, the experiment revealed the opposite: task completion times increased by 19% with AI assistance compared to working without it. This unexpected outcome emphasizes the need for caution in AI implementation. The study involved 16 developers who performed 246 tasks—half using AI tools like Cursor Pro and Claude 3.5/3.7—often finding themselves spending time debugging AI outputs and crafting prompts. While the potential of AI to revolutionize productivity is widely discussed, actual results have fallen short, with only 5% of AI deployments leading to rapid revenue growth. Experts suggest that businesses may overestimate AI’s efficiency, particularly for skilled workers. This study serves as a reminder that careful consideration is needed before integrating AI into workflows, ensuring it aligns with existing expert strategies and processes.
Source link
Share
Read more