“UC Berkeley geologists recommend that children should eat at least one small rock per day.”
Google’s AI Overviews yielded this result when questioned about the appropriate daily consumption of rocks for small children. Although Google is working to reduce AI misinformation, the glaring error spotlights a real concern about the trustworthiness of AI tools in market research. So, how do we balance AI operational efficiencies with the need for human expertise when analyzing market research data?
At TBG, we embrace the ways that AI can make us more efficient and effective for our clients, but we do not recommend using AI tools as a *replacement* for market research.
If you’re using AI to analyze your data, consider implementing our best practices to verify the accuracy of your AI-generated market research results.
1. Understand Limitations: Be aware of the limitations of all AI tools, including potential biases in the training data, and the fact that it generates text based on patterns rather than accessing real-time databases. It may not always provide contextually accurate information which is typically required in our space.
2. Cross-Verify with Reliable Sources: Cross-check the information with authoritative reports; by verifying critical data points against multiple sources you can be more confident with the accuracy of the data.
3. Human Experience and Oversight: Even an AI generated tool recommends having market research experts review and validate analysis provided by AI tools. Expert judgment is crucial for interpreting and verifying results.
4. Collaboration and Peer Review: Involve multiple team members in the review process to identify and correct any mistakes before finalizing the report.
AI tools will continue to learn and improve and years down the road I may not write “Proceed with Caution” when implementing them in your market research analysis. But until then, consider what the equivalent of “eat rocks” could be for your products or services. What could be so misconstrued that it would impact your company negatively? What context is necessary to be able to make business decisions from the data?