Productivity Improvement from Generative AI

A new study by Stanford found that generative AI assistants can significantly boost agent productivity in call centers. The AI tool helped agents resolve 13.8% more issues per hour and improved the performance of less experienced workers the most. The biggest gain was seen for agents with two months of tenure who were able to perform as well as agents with six months of tenure without the AI assistant. The researchers believe this is because the AI shares the knowledge of more experienced workers, helping new agents learn faster. However, the study also found that highly skilled workers saw little benefit from the AI as their knowledge was already captured in the AI's recommendations.

Another study from HBR recommends a cautious approach to implementing LLMs. While they may be useful for some tasks, organizations should carefully consider the potential risks before deploying them widely. They may not be the productivity boosters they promise as they

• Focus on tasks, not companies: Studies assess productivity on individual tasks, not how these tools affect entire companies. There could be unforeseen consequences.
• LLMs can be confidently wrong: LLMs are good at mimicking human language, but not necessarily factual accuracy. This can lead to problems in tasks that require truthful information.
• Can hurt top performers: LLMs trained on data from high performers can make those performers less necessary, reducing their motivation and potentially leading them to leave.
• Difficulties with retraining: As LLMs are updated with their own outputs, their quality can degrade. Additionally, retraining them requires human input, and identifying when retraining is necessary can be difficult.
• Can amplify biases: LLMs can inherit and amplify biases from the data they are trained on. This can have negative effects on company culture and employee morale.

A new study from MIT suggests that generative AI can significantly improve the performance of highly skilled workers by up to 40% if used within its(AI) capabilities. However, using AI outside its boundaries can lead to a decrease in performance of up to 19%.

Here are some key takeaways from the study:

• AI can be a powerful tool for highly skilled workers. When used for tasks that fit within its capabilities, AI can significantly boost a worker's performance.
• There's a limit to what AI can do. It's crucial to understand the boundaries of AI's abilities to avoid negative impacts on performance.
• Workers need to be aware of the limitations of AI. Over-reliance on AI recommendations can lead to incorrect conclusions.
• Human expertise remains essential. Even when using AI, skilled workers need to use their judgment and critical thinking to validate AI outputs.

The study recommends that organizations carefully consider these factors when introducing generative AI to their workforce.