As we increasingly rely on artificial intelligence (AI) to provide information, concerns about the accuracy of these sources are growing. A recent analysis by Oumi, an AI start-up, found that Google’s AI-generated answers, known as AI Overviews, were accurate approximately nine out of 10 times. However, this means that tens of millions of erroneous answers are provided every hour, or hundreds of thousands per minute.
The Power of Context
When Stephen Punwasi, a data analyst from Toronto, asked Google when Hulk Hogan had died, the answer he received was confusing: “There are no credible reports of Hulk Hogan being deceased.” This response contradicted an article from The Daily Mail that read, “Mystery Deepens Over Hulk Hogan’s Death.” Such inconsistencies highlight the importance of context in understanding AI-generated answers.
The Uncanny Valley
Google’s AI Overviews have transformed the search engine into a publisher, with AI-generated answers taking prime placement at the top of search results pages. While this new product has improved accuracy in recent months, some technologists worry that the average person may not realize those results need double-checking.
A Tale of Two Museums
Bob Marley’s home was converted into a museum in 1987, but according to Google’s AI Overviews, this occurred without any mention of the specific date. However, further research reveals that the museum actually opened on May 11, 1986, marking the fifth anniversary of Mr. Marley’s death.
The Verdict
Pratik Verma, CEO of Okahu, a company that helps people understand and use AI technologies, urges users to double-check information: “Never trust one source. Always compare what you get with another source.” Google acknowledges the possibility of errors in its AI Overviews, urging users to verify responses.
The Future of Information
As we rely more heavily on AI-generated answers, it is essential to acknowledge both their strengths and limitations. While AI can process vast amounts of information quickly and efficiently, it is crucial to recognize that these systems are not infallible. By understanding the potential for errors and taking steps to verify information, we can harness the power of AI while maintaining trust in the accuracy of our sources.