Harvard Study: Relying on AI Can Make Workers 19% More Likely to Make Mistakes

A recent study by Harvard University and Boston Consulting Group (BCG) sheds light on the impact of AI in the workplace, revealing its potential to both enhance and negatively affect productivity and efficiency. The study found that AI can enhance performance, but also that overreliance greatly increases the chances of finding wrong answers.

Forbes reports that AI is changing how we work, but it’s not all smooth sailing, according to a study by Harvard and BCG. While AI, especially OpenAI’s GPT-4, can make us work faster and better, it also has its drawbacks.

BCG consultants using GPT-4 AI did “12.2 percent more tasks” and were “25.1 percent faster.” They also gave “over 40 percent higher quality results” compared to those not using AI. Those numbers may be impressive, but they only tell part of the story. When tasks went beyond what the AI could enhance, those using it were “19 percentage points more likely” to get things wrong. BCG notes that there is a tendency for workers to rely too heavily on AI, reporting: “Even participants who were warned about the possibility of wrong answers from the tool did not challenge its output”

This “jagged technological frontier,” as researchers call it, shows AI’s limits. It can be a game-changer within its capabilities but can lead to more errors outside of them.

Business leaders are advised to know AI’s strengths and weaknesses and needs the right training to use it effectively. It’s not just about using AI to handle many tasks at once or give quick answers. For complex issues needing a human touch, AI might not be the right tool.

Leaders need to make smart choices about bringing AI into their organizations. It’s crucial to make sure employees know when and how to use it. As the study notes, we can’t fully swap people for AI.

Read more at Forbes here.

Lucas Nolan is a reporter for Breitbart News covering issues of free speech and online censorship.