From the course: Generative AI for Business Leaders

Unlock this course with a free trial

Join today to access over 25,600 courses taught by industry experts.

AI limitations: Hallucinations, bias, and data quality

AI limitations: Hallucinations, bias, and data quality

From the course: Generative AI for Business Leaders

Start my 1-month free trial Buy for my team

AI limitations: Hallucinations, bias, and data quality

“

Generative AI is powerful, but leaders need to understand its limitations. The first limitation is accuracy. AI can produce answers that sound confident, even when they are incomplete, outdated or wrong. This is often called hallucinations. That is why AI output should be checked, especially when the work affects customers, finances, safety, or reputation. The second limitation is bias. AI systems learn from data and the data can reflect the assumptions, gaps and inequalities of the world it comes from. If this is ignored, AI can produce unfair or distorted results. The third limitation is data quality. Generative AI is only as useful as the information it can draw on. If your data is messy, incomplete, out of date, or locked away in disconnected systems, the AI will struggle to deliver reliable business value. The fourth limitation is context. A general AI tool may know a lot about the world, but it may know very little about your organization, your customers, your policies, your…

Contents