What concerns do customers have regarding AI assistants like Microsoft Copilot?

Question

Accepted Answer

Customers are increasingly frustrated by the unpredictable and unreliable outputs of AI assistants like Microsoft Copilot.

Customers express significant frustration with the "meanderings" of AI assistants such as Microsoft Copilot, a sentiment highlighted by Salesforce CEO Marc Benioff. The core concern for corporate clients lies in the potential for these AI systems to "go off the rails," even if only occasionally. This unpredictability can lead to inaccurate information or unexpected behaviour, which businesses cannot afford given their operational requirements and the need for reliable outputs.
Unpredictable Outputs Cause Frustration

The primary issue for customers is that generative AI models, including Microsoft Copilot, still "hallucinate" or invent answers. This tendency to produce incorrect or fabricated information leads to a lack of trust and considerable frustration among users. The expectation of an AI assistant is to provide accurate and dependable assistance, and when it deviates from this, it undermines its utility and value in a professional setting.
Business Operations Cannot Tolerate Unreliability

For corporate customers, the risk of AI systems deviating from expected performance is a critical concern. Businesses rely on AI tools to enhance productivity and efficiency, and any instance of an AI "going off the rails" can disrupt operations, lead to financial losses, or negatively impact customer experiences. This unreliability means that even occasional errors are unacceptable in environments where precision and consistency are paramount.
The Need for AI to Admit Uncertainty

A significant gap identified by Microsoft's VP, Vik Singh, is the current inability of AI models to acknowledge when they lack sufficient information or certainty. The ideal scenario, as described, would be for the AI to raise its hand and state, "I'm not sure, I need help." This admission of uncertainty is crucial for building more robust and trustworthy AI systems that can effectively collaborate with human oversight when necessary.
Potential for Cost Savings Hinges on Reliability

Despite the current limitations, there is a clear recognition of the potential for substantial cost savings. For instance, one Microsoft client reportedly spends $8 per customer service query. If an AI assistant can handle these requests more efficiently, even if it requires human intervention in half the cases, it represents significant financial savings and improves customer response times. However, achieving these savings is contingent upon the AI's ability to provide reliable and accurate assistance.

Reading history

Hide →

This history is cleared when the page is closed.

What concerns do customers have regarding AI assistants like Microsoft Copilot?

Loading

Was this article helpful?

Helpful

Not helpful

Report an issue

1

0

0

Share with friends