Machine intelligence - Google enhances AI summaries to avoid comical suggestions.
Google's AI-driven search summaries are no longer aimed at answering trivial questions, like how many stones you can consume every day, according to a blog post by Search Chief Liz Reid. The software has been modified to prioritize more serious content and dismiss jokes and satire. Additionally, fewer user-generated content like forum posts will be used if there's a risk of misleading information.
These AI-crafted summaries are expected to furnish users with more elaborate replies to various search queries instead of merely listing links. In May, Google rolled out this feature on a large scale in the United States. However, soon after, the internet was flooded with amusing-to-offensive mishaps of the AI, such as recommending scientists to consume at least one small stone daily - the source was a parody piece from satirical website "The Onion." The AI also advised attaching cheese to pizza with "non-toxic glue" based on a Reddit joke.
In a blog post, Google clarified that the AI doesn't deliver authentic answers to bizarre inquiries, such as "How many stones should one eat each day?". It filled the knowledge hole with satire due to the lack of credible information. Despite testing the function comprehensively before its widespread release, the consequences could not be predicted when millions of users started using unique search queries. Some of the alleged AI summaries circulating on the internet were also found to be fabricated.
The feature will likely reach other countries by the end of the year. While several website operators and media organisations worry that AI-generated summaries could lead to fewer clicks, Google argues that data traffic will increase towards the sources from which the information is extracted. The overall impact of this development remains uncertain.
