Google recently introduced a new AI-powered feature in its search engine called AI Overviews, designed to answer users' complex queries by summarizing text from relevant web pages. However, this new tool immediately faced criticism and ridicule online. Despite inherent issues with generative AI that big tech companies are still struggling to resolve, Google remains committed to using AI in search results.
The AI Overviews feature, launched earlier this month, aims to provide users with concise answers to their complicated questions. However, many website owners have criticized it for essentially "stealing content" by summarizing information from their pages. Additionally, numerous users found that the tool often delivers wildly incorrect answers, which quickly spread across social media.
Google has provided several explanations for these errors. The company claims that some of the examples circulating online are fake. Others, according to Google, are due to problems such as nonsensical questions, a lack of reliable information, or the AI's difficulty in recognizing sarcasm and satire. Notably, Google did not address the possibility that rushing the product to market before it was fully ready might be the underlying issue.
A widely shared example of the AI’s mistakes involved it suggesting that people should eat rocks. Since it's unlikely anyone would seriously ask this question, the AI found an article from the satire website The Onion, which jokingly claimed that eating rocks is healthy. The AI tried to create a coherent answer based on this satirical content, leading to the absurd recommendation.
To address these problems, Google is working on improving the restrictions on the types of questions the AI will answer and where it sources its information. The AI is being adjusted to avoid spoof websites and user-generated content, and it won't generate responses to ridiculous queries in the future.
You can also read: How to Disable AI Overviews on Google Search here.
Google's explanations may account for some of the user complaints, but there are many instances shared online where the AI fails to correctly answer reasonable questions. Google argues that the embarrassing examples seen online are a loud minority, suggesting that users are less likely to discuss a tool when it works correctly. This might be a fair point, as there have been success stories shared online due to the novelty of generative AI tools like ChatGPT.
Despite claims that AI Overviews might drive traffic away from original websites by reprinting their content in the search results, Google insists that it actually drives more clicks towards those websites. The company asserts that users who find a page through AI Overviews tend to spend more time there, which they describe as "higher quality" clicks. However, these claims will likely need to be verified by third-party web traffic metrics to see if they hold true.
Overall, while Google's AI Overviews feature aims to enhance the search experience by providing quick, summarized answers, it has faced significant challenges and criticism since its launch. Google is working on refining the tool, but only time will tell if these improvements will adequately address the concerns raised by users and website owners.