A recent artificial intelligence feature called “AI Overview,” unveiled by search monolith Google, has been handing out inaccurate and dangerous summaries in response to user searches and Google doesn’t appear to have an actual fix for the problem.
As of the time of this article’s writing, Google has disabled certain queries for its “AI Overview” feature after it was widely reported that the system was generating erroneous and potentially harmful outputs.
Reports began to circulate throughout the social and news media communities of a user query asking the search engine how to keep cheese on pizza to which the AI system reportedly responded with text indicating that the user should use glue. In another batch of apparent mess ups, the AI system purportedly told users that at least two dogs owned hotels and pointed to a non-existent dog statue as evidence.
While many of the supposed inaccurate results appear humorous or innocuous, the main concerns appear to be that the consumer-facing model that generates the “AI Overview” content produces inaccurate and accurate results with the same outward confidence.
And, so far, according to Google representative Meghann Farnsworth, who spoke to the Verge via email, the company’s been relegated to removing queries that trigger inaccurate results from the system as they crop up. Essentially, it appears as though Google is playing metaphorical whack-a-mole with its AI problem.
Confusing matters even further, Google appears to be laying the blame for the issues on the humans generating the queries.
Per Farnsworth:
“Many of the examples we’ve seen have been uncommon queries, and we’ve also seen examples that were doctored or that we couldn’t reproduce.”
It’s unclear at this time how users are supposed to avoid making “uncommon queries” and, as is common with large language models, Google’s AI system has the propensity to output different answers to the same questions when queried multiple times.
Cointelegraph reached out to Google for further clarification but didn’t receive an immediate response.
While it appears as though Google’s AI system still needs some development in order to work out the kinks, the founder of rival AI company xAI, Elon Musk, believes these machines will surpass human capabilities before the end of 2025.
As Cointelegraph recently reported, Musk recently told conference-goers at the VivaTech 2024 event in Paris that he believes xAI could catch up to OpenAI and DeepMind Google by the end of 2024.
Related: Political correctness in AI systems is the biggest concern: Elon Musk