Google’s new AI Overview product has been making headlines for all the wrong reasons. Social media is filled with examples of the AI saying strange and nonsensical things, like telling users to put glue on their pizza or suggesting they eat rocks. The situation has become so chaotic that Google is scrambling to manually disable AI Overviews for specific searches as various memes flood social networks. This messy rollout is tarnishing Google’s reputation, a company once known for cutting-edge technology and high-quality output.
Despite being in testing for a year and launching in beta as the Search Generative Experience in May 2023, Google’s AI Overview product seems to have fallen short. CEO Sundar Pichai boasted about serving over a billion queries and reducing the cost of delivering AI answers by 80 percent. However, this optimization may have come too soon, before the technology was truly ready. The result is a product that is generating low-quality output that is becoming the subject of memes and ridicule.
AI expert Gary Marcus and Meta’s AI chief Yann LeCun both emphasize the challenge of achieving complete accuracy in AI systems. While getting to 80 percent correctness is relatively straightforward by approximating human data, the final 20 percent is where the real challenge lies. This last 20 percent involves complex reasoning, fact-checking, and source verification that current large language models struggle to perform. Achieving artificial general intelligence (AGI) requires a level of reasoning and critical thinking that goes beyond the capabilities of current AI systems.
Google is facing increasing competition in the AI space. Bing made headlines with its focus on AI before Google, OpenAI is reportedly working on its own search engine, and a new AI search startup is already valued at $1 billion. The pressure to innovate and stay ahead of the curve has led to messy AI releases and failed product launches, reminiscent of Meta’s Galactica AI system that had to be taken down for suggesting dangerous actions. The younger generation of users, who prioritize user experience, are also switching to platforms like TikTok, posing a threat to Google’s dominance.
Google has ambitious plans for its AI Overview product beyond the current chaos. The company aims to implement multistep reasoning for complex queries, generate AI-organized results pages, and introduce video search capabilities in Google Lens. However, before diving into these advanced features, Google must focus on getting the basics right. The reputation of the company is at stake, as users expect accurate and reliable information from AI systems. It’s clear that the industry as a whole needs to prioritize sanity checking and critical thinking capabilities in AI systems to avoid embarrassing and potentially harmful outputs.
Leave a Reply