"The AI takeover of Google Search starts now," writes The Verge's David Pierce. At Google I/O today, the company demoed a new opt-in feature called Search Generative Experience (SGE). The new experience generates AI "snapshots" that appear at the top of the search results page consisting of an AI-generated summary about your query, with links to sources of information and shopping. From the report: To demonstrate, Liz Reid, Google's VP of Search, flips open her laptop and starts typing into the Google search box. "Why is sourdough bread still so popular?" she writes and hits enter. Google's normal search results load almost immediately. Above them, a rectangular orange section pulses and glows and shows the phrase "Generative AI is experimental." A few seconds later, the glowing is replaced by an AI-generated summary: a few paragraphs detailing how good sourdough tastes, the upsides of its prebiotic abilities, and more. To the right, there are three links to sites with information that Reid says "corroborates" what's in the summary.

Google calls this the "AI snapshot." All of it is by Google's large language models, all of it sourced from the open web. Reid then mouses up to the top right of the box and clicks an icon Google's designers call "the bear claw," which looks like a hamburger menu with a vertical line to the left. The bear claw opens a new view: the AI snapshot is now split sentence by sentence, with links underneath to the sources of the information for that specific sentence. This, Reid points out again, is corroboration. And she says it's key to the way Google's AI implementation is different. "We want [the LLM], when it says something, to tell us as part of its goal: what are some sources to read more about that?"

A few seconds later, Reid clicks back and starts another search. This time, she searches for the best Bluetooth speakers for the beach. Again, standard search results appear almost immediately, and again, AI results are generated a few seconds later. This time, there's a short summary at the top detailing what you should care about in such a speaker: battery life, water resistance, sound quality. Links to three buying guides sit off to the right, and below are shopping links for a half-dozen good options, each with an AI-generated summary next to it. I ask Reid to follow up with the phrase "under $100," and she does so. The snapshot regenerates with new summaries and new picks. "This is the new look of Google's search results page," concludes Pierce. "It's AI-first, it's colorful, and it's nothing like you're used to. It's powered by some of Google's most advanced LLM work to date, including a new general-purpose model called PaLM 2 and the Multitask Unified Model (MUM) that Google uses to understand multiple types of media."

"In the demos I saw, it's often extremely impressive. And it changes the way you'll experience search, especially on mobile, where that AI snapshot often eats up the entire first page of your results."

DuckDuckGo announced a new tool called DuckAssist that "automatically pulls and summarizes information from Wikipedia in response to certain questions," reports The Verge. From the report: DuckAssist's beta is live on the search engine right now -- but only through DuckDuckGo's mobile apps and browser extensions. Gabriel Weinberg, the founder and CEO of DuckDuckGo, says the company will add it to the web-based search engine if the trial "goes well." When you enter a question that DuckAssist can help with, you'll see a box that says "I can check to see if Wikipedia has relevant info on this topic, just ask" at the very top of your search results. Hit the blue "Ask" button, and you'll get an AI-generated answer using summarized information from Wikipedia. If DuckAssist has already answered a question once before, that response will automatically appear, which means you won't have to "ask" it the same thing multiple times.

While the tool's built upon language models from OpenAI, the company that makes ChatGPT, and the Google-backed Anthropic, Weinberg says it'll retain the same focus on privacy as DuckDuckGo. According to the announcement, DuckAssist won't share any personally identifiable information with OpenAI and Anthropic, and neither company will use your anonymous questions to train their models. DuckDuckGo says the feature uses the "most recent full Wikipedia download available," which is around a few weeks old, so it might not be able to help if you're searching for something later than that. However, the company plans to update this in the future, as well as add more sources for DuckAssist to draw from.

Microsoft today announced that its new AI-enabled Bing will now allow users to generate images with Bing Chat. From a report: This new feature is powered by DALL-E, OpenAI's generative image generator. The company didn't say which version of DALL-E it is using here, except for saying that it is using the "very latest DALL-E models." Dubbed the "Bing Image Creator," this new capability is now (slowly) rolling out to users in the Bing preview and will only be available through Bing's Creative Mode. It'll come to Bing's Balanced and Precise modes in the future.

The new image generator will also be available in the Edge sidebar. The right prompts will generate the now-familiar square of four high-res DALL-E images. There's one major difference, though: there will be a small Bing logo in the bottom left corner. The early Bing AI release was missing a few guardrails, but Microsoft quickly fixed those. The company is clearly hoping to avoid these issues with this release.

Thomas Claburn writes via The Register: GitHub has a lot of code to search -- more than 200 million repositories -- and says last November's beta version of a search engine optimized for source code that has caused a "flurry of innovation." GitHub engineer Timothy Clem explained that the company has had problems getting existing technology to work well. "The truth is from Solr to Elasticsearch, we haven't had a lot of luck using general text search products to power code search," he said in a GitHub Universe video presentation. "The user experience is poor. It's very, very expensive to host and it's slow to index." In a blog post on Monday, Clem delved into the technology used to scour just a quarter of those repos, a code search engine built in Rust called Blackbird.

Blackbird currently provides access to almost 45 million GitHub repositories, which together amount to 115TB of code and 15.5 billion documents. Shifting through that many lines of code requires something stronger than grep, a common command line tool on Unix-like systems for searching through text data. Using ripgrep on an 8-core Intel CPU to run an exhaustive regular expression query on a 13GB file in memory, Clem explained, takes about 2.769 seconds, or 0.6GB/sec/core. [...] At 0.01 queries per second, grep was not an option. So GitHub front-loaded much of the work into precomputed search indices. These are essentially maps of key-value pairs. This approach makes it less computationally demanding to search for document characteristics like the programming language or word sequences by using a numeric key rather than a text string. Even so, these indices are too large to fit in memory, so GitHub built iterators for each index it needed to access. According to Clem, these lazily return sorted document IDs that represent the rank of the associated document and meet the query criteria.

To keep the search index manageable, GitHub relies on sharding -- breaking the data up into multiple pieces using Git's content addressable hashing scheme and on delta encoding -- storing data differences (deltas) to reduce the data and metadata to be crawled. This works well because GitHub has a lot of redundant data (e.g. forks) -- its 115TB of data can be boiled down to 25TB through deduplication data-shaving techniques. The resulting system works much faster than grep -- 640 queries per second compared to 0.01 queries per second. And indexing occurs at a rate of about 120,000 documents per second, so processing 15.5 billion documents takes about 36 hours, or 18 for re-indexing since delta (change) indexing reduces the number of documents to be crawled.

Google is introducing a new online safety feature to help users avoid inadvertently seeing graphically violent or pornographic images while using its search engine. From a report: Announced as part of the company's Safer Internet Day event on Tuesday, the new default setting enabled for everyone will automatically blur explicit images that appear in search results, even for users that don't have SafeSearch enabled. Google has confirmed to The Verge that, should they wish, signed-in users over 18 will be able to disable the blur setting entirely after it launches in "the coming months."

People in Europe can get Google to delete search results about them if they prove the information is "manifestly inaccurate," the EU's top court has ruled. From a report: The case kicked off when two investment managers requested Google to dereference results of a search made on the basis of their names, which provided links to certain articles criticising that group's investment model. They say those articles contain inaccurate claims. Google refused to comply, arguing that it was unaware whether the information contained in the articles was accurate or not. But in a ruling Thursday, the Court of Justice of the European Union opened the door to the investment managers being able to successfully trigger the so-called "right to be forgotten" under the EU's General Data Protection Regulation. "The right to freedom of expression and information cannot be taken into account where, at the very least, a part -- which is not of minor importance -- of the information found in the referenced content proves to be inaccurate," the court said in a press release accompanying the ruling.

Google's search results on desktop will load in a continuous scroll instead of dividing into pages, the company has announced. From a report: The move follows a similar change made on mobile in October last year, but isn't quite an "infinite" scroll. Instead, Google will load six pages of results into a single scroll before offering users a "See more" button to show more results. Google says the change is rolling out first for English searches in the US, but judging by the rollout of the feature on mobile it seems safe to expect to see additional markets and languages added over time.

Google says "60% of the internet is duplicate." Gary Illyes from Google posted this slide at the Google Search Central Live in Singapore the other day.

Google announced today that it's introducing a slew of new Maps, Search and Shopping features. The company revealed a majority of the new features during its Search On event in September and is now starting to roll them out to users. TechCrunch reports: Search
Starting today, users will be able to use Search to find their favorite dish at a restaurant near them. For example, you can search "truffle mac and cheese near me" to see which nearby restaurants carry the dish on their menu. Once you find a specific dish that you're looking for, you can get more information about its price, ingredients and more. Another new Search functionality lets you use Google's multisearch feature to find specific food near you. Say you see something tasty-looking online, but don't know what it is or where to find it. You can now use Lens in the Google app for Android or iOS to snap a picture or take a screenshot of a dish and add the words "near me" to find a place that sells it nearby. Later this year, Google is going roll out an update to its Lens AR Translate capabilities so users can more seamlessly translate text on complex backgrounds. Instead of covering up the original text like it currently does, Google is going to erase the text and re-create the pixels underneath with an AI-generated background, and then overlay the translated text on top of the image.

Maps
As for the new Maps features, Google is launching a new visual search experience called Live View in London, Los Angeles, New York, Paris, San Francisco and Tokyo. [...] In addition to displaying information about where places are, users will be able to see key information about each spot overlaid, such as whether the location is busy, if its open, what the price range is, etc. Another new Maps feature makes it easier for EV owners to find the best charging station for their vehicle. Now, you can search for "EV charging stations" and select the "fast charge" filter. You can also filter for stations that offer your EV's plug type. Google also announced that it's expanding its "accessible places" feature globally after initially launching it in the U.S., Australia, Japan and the U.K. in 2020. The feature is designed to help people determine whether a place is wheelchair accessible.

Shopping
Google has announced a new AR shopping feature that is designed to make it easier to find your exact foundation match. The company says its new photo library features 148 models representing a diverse spectrum of skin tones, ages, genders, face shapes, ethnicities and skin types. As a result, it should be easier for shoppers to better visualize what different products will look like on them. [...] Users can now also shop for shoes using AR.

Four years after Laserlike was acquired by Apple to boost its web search technology, the founders have quit to rejoin Google. AppleInsider reports: Prior to forming Laserlike in 2015, Anand Shukla, Srinivasan Venkatachary and Steven Baker were all Google employees. Their work at Apple is one reason the company has been predicted to launch its own search engine equivalent to Google's. Apple acquired Laserlike in 2018, though the deal wasn't made public until the following year. As part of the acquisition, the company's three founders were, and ultimately they led a 200-strong search team at Apple.

According to The Information, Srinivasan Venkatachary has now returned to Google. He is reportedly the company's new vice president of engineering. Venkatachary reports to James Manyika, senior vice president of technology and society. Baker and Shukla now both work on Manyika's team. It's not known whether all three quit Apple at the same time, or whether Venkatachary is just the latest to move. According to The Information's source, Apple is estimated to be about four years away from launching a Google Search rival.

Google is making it easier to find search results from Reddit and other forum sites. Engadget reports: The search engine is adding a new module that will surface discussions happening on forums across the web for queries that may benefit from crowd-sourced answers. The "discussions and forums" module will surface relevant posts from sites like Reddit and Quora alongside more traditional search results. It's not clear exactly how Google is determining what types of searches are best suited to forum posts. The company says the new "forum" results will "appear when you search for something that might benefit from the diverse personal experiences found in online discussions."

Google is also adding a new feature to news-related searches that will make it easier to browse international headlines that are published in languages other than English. With the change, news-related searches will also turn up relevant local coverage translated by Google. In other Google Search-related news, the company announced that starting today people in the U.S. will be able to use their new "Results About You" feature, "which aims to provide a simpler way for people to get their sensitive personal information out of the company's search results," reports Gizmodo. "Next year, Results About You will become proactive and allow users to opt in to alerts when new personal information related to them appears in search results, enabling users to request removal more quickly."

Alphabet's Google pays billions of dollars each year to Apple, Samsung Electronics and other telecom giants to illegally maintain its spot as the No. 1 search engine, the US Justice Department told a federal judge Thursday. From a report: DOJ attorney Kenneth Dintzer didn't disclose how much Google spends to be the default search engine on most browsers and all US mobile phones, but described the payments as "enormous numbers."

"Google invests billions in defaults, knowing people won't change them," Dintzer told Judge Amit Mehta during a hearing in Washington that marked the first major face-off in the case and drew top DOJ antitrust officials and Nebraska's attorney general among the spectators. "They are buying default exclusivity because defaults matter a lot." Google's contracts form the basis of the DOJ's landmark antitrust lawsuit, which alleges the company has sought to maintain its online search monopoly in violation of antitrust laws. State attorneys general are pursuing a parallel antitrust suit against the search giant, also pending before Mehta.

ByteDance, owner of the hit short video app TikTok, has quietly launched a new search engine that promises no advertisements in a cyberspace where Google has not been available for more than a decade. From a report: Without any announcement, ByteDance subsidiary Beijing Infinite Dimension Technology launched the Wukong search app this month, within days of Tencent Holdings shutting down on August 8 its Sogou search app. Sogou, which Tencent bought last year, still maintains its web-based search engine. Wukong, currently available on Apple's App Store in China and various Chinese Android app stores, brings ByteDance into closer competition with Baidu, China's dominant search engine. The new app promotes itself as providing "quality information and search without ads." The line could be interpreted as an indirect jab at Baidu, which has long faced criticism for its paid listings in search results. In 2016, 21-year-old college student Wei Zexi died of a rare cancer after he received experimental treatments recommended by Baidu.

Google announced today that it's rolling out new Search updates over the next few weeks that will aim to make it easier for people to find high-quality content. TechCrunch reports: The new ranking improvements will work to reduce the amount of low-quality or unoriginal content that ranks high in search results. Google says that the update will especially target content that has been created primarily for ranking on search engines, known as "SEO-first" content, rather than human-first content. The company's tests have shown that the update will improve the results users find when searching for content like online educational materials, as well as arts and entertainment, shopping and tech-related content.

The new updates should help reduce the number of low-quality results from websites that have learned to game the system with content that is optimized to rank high in search results. Google says users should start to see content that is actually useful rank more prominently in search results. The company plans to refine its systems and build on these improvements over time. "If you search for information about a new movie, you might have previously encountered articles that aggregated reviews from other sites without adding perspectives beyond what's available elsewhere on the web," the company explained in a blog post. "This isn't very helpful if you're expecting to read something new. With this update, you'll see more results with unique information, so you're more likely to read something you haven't seen before."

Google will stop giving snappy answers to stupid questions, the company has announced, as it seeks to improve its search engine's "featured snippets" service. From a report: That means users should see fewer answers to questions such as "When did Snoopy assassinate Abraham Lincoln?", to which the service would once merrily respond with "1865" -- the right date, but very much the wrong assassin. "This clearly isn't the most helpful way to display this result," said the company's head of search, Pandu Nayak, in a blogpost announcing the changes. "We've trained our systems to get better at detecting these sorts of false premises, which are not very common, but there are cases where it's not helpful to show a featured snippet. We've reduced the triggering of featured snippets in these cases by 40% with this update."

Snippets, which sometimes show up as a featured response to direct questions asked of Google Search, have long been a cornerstone of the company's AI strategy. The same technology powers its smart speakers and voice assistants, and lets the search engine satisfy search queries without visitors clicking away to other websites. But the snippets, which are automatically generated from the contents of websites, have also been a thorn in Google's side for just as long. [...] In an effort to address the root cause of such mistakes, Google is also rolling out new warnings for times when a search term has hit a "data void" -- a question where a good answer might simply not exist.

Google is still useful for many, but the harder question is why its results feel more sterile than they did five years ago. From a report: SEO expert Marie Haynes's theory is that this is the result of Google trying to crack down on misinformation and low-quality content -- especially around consequential search topics. In 2017, the company started talking publicly about a Search initiative called EAT, which stands for "expertise, authoritativeness, and trustworthiness." The company has rolled out numerous quality rater guidelines, which help judge content to determine authenticity. One such effort, titled Your Money or Your Life, applies rigorous standards to any pages that show up when users search for medical or financial information.

"Take crypto," Haynes explained. "It's an area with a lot of fraud, so unless a site has a big presence around the web and Google gets the sense they're known for expertise on that topic, it'll be difficult to get them to rank." What this means, though, is that Google's results on any topic deemed sensitive enough will likely be from established sources. Medical queries are far more likely to return WebMD or Mayo Clinic pages, instead of personal testimonials. This, Haynes said, is especially challenging for people looking for homeopathic or alternative-medicine remedies.

There's a strange irony to all of this. For years, researchers, technologists, politicians, and journalists have agonized and cautioned against the wildness of the internet and its penchant for amplifying conspiracy theories, divisive subject matter, and flat-out false information. Many people, myself included, have argued for platforms to surface quality, authoritative information above all else, even at the expense of profit. And it's possible that Google has, in some sense, listened (albeit after far too much inaction) and, maybe, partly succeeded in showing higher-quality results in a number of contentious categories. But instead of ushering in an era of perfect information, the changes might be behind the complainers' sense that Google Search has stopped delivering interesting results.

Alphabet's Google unveiled a series of planned upgrades to its search and maps services revealing the company's augmented reality ambitions -- and its appeal to a generation of internet users drifting away from the company. From a report: The new features include ways for people to search for nearby items using images and identify physical objects with their smartphone cameras. On Google Maps, the company promised a way for people to explore detailed 3D digital models of landmarks and neighborhoods before setting foot in person. Google shared the plans on Wednesday for the first day of its annual I/O developer conference held near its Mountain View, California headquarters.

Google is working to keep its products relevant and growing as users' needs evolve beyond text. "Search should be something that you can do anywhere, in any way you want, using any of your senses," Prabhakar Raghavan, Google's senior vice president and product chief, said in an interview. Google's core search advertising business has continued to grow steadily during the pandemic, despite recent middling financial results. Yet the I/O announcements underscored nascent threats Google sees to its flagship services. People in emerging markets are more likely to search with voice features than typing, which has driven Google to invest more in its voice assistant feature.

DuckDuckGo, a privacy-oriented search engine, netted more than 35 billion search queries in 2021, a 46.4% jump over 2020 (23.6 billion). From a report: That's big. Even so, the company, which bills itself as the "Internet privacy company," offering a search engine and other products designed to "empower you to seamlessly take control of your personal information online without any tradeoffs," remains a rounding error compared to Google in search. Whether it remains a highly successful (and profitable) rounding error, however, could depend on how serious we become about the privacy of our searches.

Google has announced that it's changing the way search works on mobile devices, initially in the U.S. From a report: Now, when you reach the bottom of a set of search results on your phone, you won't have to tap to go to the next page. Instead, the next set of results will automatically load so you can continuously scroll down to see more information. The change will roll out on the mobile web and will be supported on the Google mobile app for both iOS and Android in the U.S. for most English-language searches for the time being. Because it's a staggered release, you may initially encounter some results which scroll and others that do not.

At its Search On event today, Google introduced several new features that, taken together, are its strongest attempts yet to get people to do more than type a few words into a search box. From a report: By leveraging its new Multitask Unified Model (MUM) machine learning technology in small ways, the company hopes to kick off a virtuous cycle: it will provide more detail and context-rich answers, and in return it hopes users will ask more detailed and context-rich questions. The end result, the company hopes, will be a richer and deeper search experience. Google SVP Prabhakar Raghavan oversees search alongside Assistant, ads, and other products. He likes to say -- and repeated in an interview this past Sunday -- that "search is not a solved problem." That may be true, but the problems he and his team are trying to solve now have less to do with wrangling the web and more to do with adding context to what they find there.

For its part, Google is going to begin flexing its ability to recognize constellations of related topics using machine learning and present them to you in an organized way. A coming redesign to Google search will begin showing "Things to know" boxes that send you off to different subtopics. When there's a section of a video that's relevant to the general topic -- even when the video as a whole is not -- it will send you there. Shopping results will begin to show inventory available in nearby stores, and even clothing in different styles associated with your search. For your part, Google is offering new ways to search that go beyond the text box. It's making an aggressive push to get its image recognition software Google Lens into more places. It will be built into the Google app on iOS and also the Chrome web browser on desktops. And with MUM, Google is hoping to get users to do more than just identify flowers or landmarks, but instead use Lens directly to ask questions and shop.