You can now get answers to all the stupid questions you're too embarrassed to ask others or difficult to express in a traditional Google search.
This week's Google I/O keynote was a two-hour trumpet about all the ways AI powers and permeates many of the company's biggest software and apps. A demonstration of how Gemini, Google's flagship generative AI-powered chatbot, enhances existing AI capabilities. But one of the more impressive examples is how search can answer questions asked while filming a video.
This is the future of AI that I want when I'm afraid of embarrassment when I can't figure out a seemingly obvious part of my car or whether I should see a doctor for a rash.
On the other hand, we can't ignore how much the quality of Google search has declined over the past few years, amplifying its usefulness. The company has invented an effective Band-Aid for a problem that has continued to worsen.
Google I/O 2024 details
On the Google I/O stage, Rose Yao, vice president of Google Search Products, showed viewers how. She used her Google Lens to troubleshoot her malfunctioning record player, recording a video while discreetly asking aloud, “Why doesn't this stay in place?” .
Mr. Yao did not name the part in question, the tone arm that carries the needle over the record, forcing Mr. Lenz to suggest an answer using context clues. Search provides an AI summary of what the problem is (tone arm balancing), provides suggestions for fixes, identifies the make and model of your record player, spotlights sources, and helps you find more information. We have made it possible for you to search for answers.
read more: Google, Project Astra, AI Overview, Gemini update ups its AI game
Yao explained that this process was made possible by stringing together a series of AI queries into a seamless procedure. Natural language processing parsed her audio requests, then Gemini's Context window broke down the video frame by frame, identified the record player, and tracked the movement of problematic parts. I then scoured her forums, articles, and videos online to find the one that best matched my query for Yao's video (in this case, her Audio-Technica article, an audiophile manufacturer).
For now, you can do all of these things individually and end up arriving at more or less the same answer. You can point Google Lens at something and have it identify the object. You can also word your problem carefully and hope someone asks a similar question on Quora, Reddit, etc. You can also narrow down your search by searching for the brand of your record player and playing around with how to find your exact model.
But assuming Google Lens with Gemini works as demonstrated, you won't be able to answer questions on the internet as fast as we saw on the Google I/O stage. Perhaps more importantly, you'll get lots of help with asking sensitive and sometimes embarrassing questions.
Let's consider the possibilities. You might ask, “What part of the car is this?” “How often should I change them?” you might say, pointing to your bed sheets. “What's the best way to clean this?” It's like sitting in your car and pointing at a food stain on your shirt. “How do I turn this into a pitcher of margaritas?” you might ask, feeling overconfident and pointing at the counter piled with ingredients. Or you might point to a shaped part of your body that bothers you and ask, “Should I have this checked?”
Google Lens, Search, and its AI tools are not a substitute for expert knowledge or medical perspective, so don't think that the company can replace expert opinion. But it can help you get over the initial painful hurdle of trying to figure it out. what search for. In the record player example above, I needed a text to explain which part was the problem. That's why I searched for “record player anatomy” to visually identify that part while writing this article.
Experienced internet searchers can search from there. However, Google Lens can quickly eliminate the friction of narrowing down your search when troubleshooting a specific issue. This can be even more difficult if this is a rare problem with sparse results. If your search terms make it difficult to pinpoint the problem, and you experience a combination of frustration and embarrassment, you may abandon your search.
So, assuming the Google Lens process works broadly enough that people can use it to look at things in real life, it's an excellent solution to many simple questions that might not have been answered a few decades ago. It seems like a good way to do it. For people with severe anxiety, asking for help from her faceless Google Lens instead of a human can be a lifesaver.
And even better if you can use Google Lens to ask which part of your engine is the oil cap without having to worry about the judgment of your longtime mechanic.
Of course, these answers are only useful if they are correct. A Google I/O promotional video shared with viewers had another example of using Google Lens to get answers. In this case, the film camera is malfunctioning. As The Verge noticed, the answer provided by Search's AI involved opening the back plate, which exposed it to sunlight and potentially ruined the undeveloped roll of film. there was.
If a company's AI cannot avoid harmful suggestions, it should not casually parse online information sources. Then again, maybe the reason I'm so intrigued by the search results generated by AI is because it's getting harder to find useful information online.
AI, the Band-Aid of Google Search
A new and useful feature in Google Lens is a reminder that finding information on the internet can be difficult these days. Search results are fronted with ads that look like legitimate links, highlighted in results after multiple algorithm tweaks over the years have mixed up the first results. The overall quality of the sites displayed seems to be much worse than before.
As sites look to SEO techniques to rank their pages higher than their competitors, search The ecosystem is in trouble (full disclosure: CNET uses several SEO techniques). I've heard more than one friend ruefully say that he adds “Reddit” to all his Google searches just to have a chance of getting his queries answered.
In this reality, manual searches yield less useful results year after year, so using AI to automatically parse the nonsense seems like a better choice. But for the search ecosystem, this seems like a temporary fix that will be harmful in the long run. If enough people rely on AI to perform searches, the sites that rely on that traffic will dry up and there will be no online answers that Google can send to his AI to retrieve.
Editor's note: CNET used an AI engine to create dozens of stories and label them accordingly. The notes you are reading are attached to articles that substantively cover the topic of AI, all written by professional editors and writers. Learn more about. AI policy.