Google has made it clear that it should not hold back and push ahead with its recent artificial intelligence (AI)-based search even when problems arise. The reason is that the introduction of AI-based search has turn into an irreversible trend.
CNBC obtained an internal audio file on the thirteenth (local time), during which Liz Reed, Vice President of Google Search, said in a general meeting, “Mistakes in AI products are inevitable, but the corporate must proceed to push forward and resolve problems as they arise.” reported the facts.
Reid acknowledged that errors are sure to occur as AI becomes more widespread in Google Search and other products, and emphasized the importance of taking urgent motion to resolve problems after they are discovered reasonably than taking risks. . Which means that AI searches won’t be placed on hold on account of potential problems.
Google has recently been criticized for its ‘AI Overview’ tool, which provides users with meaningless answers in searches. The AI overview tool provided false information, including claiming that Barack Obama was America’s first Muslim president and suggesting users eat rocks or put glue on pizza.
Previously, Google suffered by temporarily suspending the discharge of the image creation function of ‘Gemini’, the bottom model for AI search, on account of historical inaccuracies and questionable responses.
Nonetheless, image creation and search are on a distinct level. Most of Google’s revenue comes from search-related promoting. In a situation where OpenAI’s ChatGPT is expanding its base and Microsoft (MS) is entering the search competition based on it, slowing down could cause big problems.
For that reason, Google has strengthened its measures, equivalent to correcting inaccurate and bizarre results one after the other and restricting queries that result in sensitive health issues or strange answers. He also protested that most of the user criticisms were almost unreasonable.
Reid said Google conducts extensive testing and “red teams” to discover vulnerabilities in AI products before releasing them, but acknowledged that more testing continues to be needed.
It also highlighted the difficulties in understanding the standard and context of web page content and encouraged employees to report any problems they encounter. This emphasizes the importance of worker feedback in improving AI products.
This is alleged to point out that Google’s will to just accept mistakes and improve has turn into stronger.
Reporter Park Chan cpark@aitimes.com
