"When Google added a new button for searching the web with generative artificial intelligence, the feature debuted with such glaring flaws, like a tendency to make things up, that I was skeptical it would become my go-to tool for finding information online. But one year later, I confess that I have become a convert. Even though the technology remains imperfect, I am increasingly clicking the button, labeled “AI Mode,” on Google.com to type requests and immediately finish tasks that would have required many minutes with an old-school search."
"It took me some experimenting to get better results from AI search, and the key was to tell Google to work with a small amount of information instead of crawling the web for answers. Recently, I used Google's AI search for identifying a car part to do a repair, picking out a condiment at a grocery store and spotting internet scams. With a normal web search, I would have had to do multiple keyword queries and read several articles to get the job done, but Google's AI essentially automated the process."
"A big caveat: I recommend steering clear of using AI Mode as a regular search engine, looking for direct answers to questions. Google's AI often spews misinformation; just last week, it incorrectly said a car diagnostics app could tell me whether my vehicle was ready for a smog check, a nonexistent feature. A New York Times analysis found that roughly 10% of Google's AI-generated answers were wrong."
"With Google processing more than 5 trillion searches a year, that translates to tens of millions of erroneous answers every hour; in our analysis, Google fumbled with basic facts, including historical dates, names and celebrity news. Google said the overwhelming majority of its AI responses were accurate."
Google’s generative AI search feature, labeled “AI Mode,” initially showed obvious flaws, including making up information. After experimenting, better results came from providing a small amount of relevant information rather than asking the system to crawl the web for answers. AI Mode has been used for practical tasks such as identifying a car part for a repair, choosing a condiment in a grocery store, and detecting internet scams, reducing the time and number of searches needed. A major caveat remains: AI Mode should not be relied on for direct answers, because it can produce incorrect claims, including nonexistent features. An analysis estimated about 10% of AI-generated answers were wrong, implying many errors at scale.
Read at Miami Herald
Unable to calculate read time
Collection
[
|
...
]