Google

man-vs.-machine:-deepmind’s-new-robot-serves-up-a-table-tennis-triumph

Man vs. machine: DeepMind’s new robot serves up a table tennis triumph

John Henry was a steel-driving man —

Human-beating ping-pong AI learned to play in a simulated environment.

A blue illustration of a robotic arm playing table tennis.

Benj Edwards / Google DeepMind

On Wednesday, researchers at Google DeepMind revealed the first AI-powered robotic table tennis player capable of competing at an amateur human level. The system combines an industrial robot arm called the ABB IRB 1100 and custom AI software from DeepMind. While an expert human player can still defeat the bot, the system demonstrates the potential for machines to master complex physical tasks that require split-second decision-making and adaptability.

“This is the first robot agent capable of playing a sport with humans at human level,” the researchers wrote in a preprint paper listed on arXiv. “It represents a milestone in robot learning and control.”

The unnamed robot agent (we suggest “AlphaPong”), developed by a team that includes David B. D’Ambrosio, Saminda Abeyruwan, and Laura Graesser, showed notable performance in a series of matches against human players of varying skill levels. In a study involving 29 participants, the AI-powered robot won 45 percent of its matches, demonstrating solid amateur-level play. Most notably, it achieved a 100 percent win rate against beginners and a 55 percent win rate against intermediate players, though it struggled against advanced opponents.

A Google DeepMind video of the AI agent rallying with a human table tennis player.

The physical setup consists of the aforementioned IRB 1100, a 6-degree-of-freedom robotic arm, mounted on two linear tracks, allowing it to move freely in a 2D plane. High-speed cameras track the ball’s position, while a motion-capture system monitors the human opponent’s paddle movements.

AI at the core

To create the brains that power the robotic arm, DeepMind researchers developed a two-level approach that allows the robot to execute specific table tennis techniques while adapting its strategy in real time to each opponent’s playing style. In other words, it’s adaptable enough to play any amateur human at table tennis without requiring specific per-player training.

The system’s architecture combines low-level skill controllers (neural network policies trained to execute specific table tennis techniques like forehand shots, backhand returns, or serve responses) with a high-level strategic decision-maker (a more complex AI system that analyzes the game state, adapts to the opponent’s style, and selects which low-level skill policy to activate for each incoming ball).

The researchers state that one of the key innovations of this project was the method used to train the AI models. The researchers chose a hybrid approach that used reinforcement learning in a simulated physics environment, while grounding the training data in real-world examples. This technique allowed the robot to learn from around 17,500 real-world ball trajectories—a fairly small dataset for a complex task.

A Google DeepMind video showing an illustration of how the AI agent analyzes human players.

The researchers used an iterative process to refine the robot’s skills. They started with a small dataset of human-vs-human gameplay, then let the AI loose against real opponents. Each match generated new data on ball trajectories and human strategies, which the team fed back into the simulation for further training. This process, repeated over seven cycles, allowed the robot to continuously adapt to increasingly skilled opponents and diverse play styles. By the final round, the AI had learned from over 14,000 rally balls and 3,000 serves, creating a body of table tennis knowledge that helped it bridge the gap between simulation and reality.

Interestingly, Nvidia has also been experimenting with similar simulated physics systems, such as Eureka, that allow an AI model to rapidly learn to control a robotic arm in simulated space instead of the real world (since the physics can be accelerated inside the simulation, and thousands of simultaneous trials can take place). This method is likely to dramatically reduce the time and resources needed to train robots for complex interactions in the future.

Humans enjoyed playing against it

Beyond its technical achievements, the study also explored the human experience of playing against an AI opponent. Surprisingly, even players who lost to the robot reported enjoying the experience. “Across all skill groups and win rates, players agreed that playing with the robot was ‘fun’ and ‘engaging,'” the researchers noted. This positive reception suggests potential applications for AI in sports training and entertainment.

However, the system is not without limitations. It struggles with extremely fast or high balls, has difficulty reading intense spin, and shows weaker performance in backhand plays. Google DeepMind shared an example video of the AI agent losing a point to an advanced player due to what appears to be difficulty reacting to a speedy hit, as you can see below.

A Google DeepMind video of the AI agent playing against an advanced human player.

The implications of this robotic ping-pong prodigy extend beyond the world of table tennis, according to the researchers. The techniques developed for this project could be applied to a wide range of robotic tasks that require quick reactions and adaptation to unpredictable human behavior. From manufacturing to health care (or just spanking someone with a paddle repeatedly), the potential applications seem large indeed.

The research team at Google DeepMind emphasizes that with further refinement, they believe the system could potentially compete with advanced table tennis players in the future. DeepMind is no stranger to creating AI models that can defeat human game players, including AlphaZero and AlphaGo. With this latest robot agent, it’s looking like the research company is moving beyond board games and into physical sports. Chess and Jeopardy have already fallen to AI-powered victors—perhaps table tennis is next.

Man vs. machine: DeepMind’s new robot serves up a table tennis triumph Read More »

all-the-possible-ways-to-destroy-google’s-monopoly-in-search

All the possible ways to destroy Google’s monopoly in search

All the possible ways to destroy Google’s monopoly in search

Aurich Lawson

After US District Judge Amit Mehta ruled that Google has a monopoly in two markets—general search services and general text advertising—everybody is wondering how Google might be forced to change its search business.

Specifically, the judge ruled that Google’s exclusive deals with browser and device developers secured Google’s monopoly. These so-called default agreements funneled the majority of online searches to Google search engine result pages (SERPs), where results could be found among text ads that have long generated the bulk of Google’s revenue.

At trial, Mehta’s ruling noted, it was estimated that if Google lost its most important default deal with Apple, Google “would lose around 65 percent of its revenue, even assuming that it could retain some users without the Safari default.”

Experts told Ars that disrupting these default deals is the most obvious remedy that the US Department of Justice will seek to restore competition in online search. Other remedies that may be sought range from least painful for Google (mandating choice screens in browsers and devices) to most painful (requiring Google to divest from either Chrome or Android, where it was found to be self-preferencing).

But the remedies phase of litigation may have to wait until after Google’s appeal, which experts said could take years to litigate before any remedies are ever proposed in court. Whether Google could be successful in appealing the ruling is currently being debated, with anti-monopoly advocates backing Mehta’s ruling as “rock solid” and critics suggesting that the ruling’s fresh takes on antitrust law are open to attack.

Google declined Ars’ request to comment on appropriate remedies or its plan to appeal.

Previously, Google’s president of global affairs, Kent Walker, confirmed in a statement that the tech giant would be appealing the ruling because the court found that “Google is ‘the industry’s highest quality search engine, which has earned Google the trust of hundreds of millions of daily users,’ that Google ‘has long been the best search engine, particularly on mobile devices,’ ‘has continued to innovate in search,’ and that ‘Apple and Mozilla occasionally assess Google’s search quality relative to its rivals and find Google’s to be superior.'”

“Given this, and that people are increasingly looking for information in more and more ways, we plan to appeal,” Walker said. “As this process continues, we will remain focused on making products that people find helpful and easy to use.”

But Mehta found that Google was wielding its outsize influence in the search industry to block rivals from competing by locking browsers and devices into agreements ensuring that all searches went to Google SERPs. None of the pro-competitive benefits that Google claimed justified the exclusive deals persuaded Mehta, who ruled that “importantly,” Google “exercised its monopoly power by charging supra-competitive prices for general search text ads”—and thus earned “monopoly profits.”

While experts think the appeal process will delay litigation on remedies, Google seems to think that Mehta may rule on potential remedies before Google can proceed with its appeal. Walker told Google employees that a ruling on remedies may arrive in the next few months, The Wall Street Journal reported. Ars will continue monitoring for updates on this timeline.

As the DOJ’s case against Google’s search business has dragged on, reports have long suggested that a loss for Google could change the way that nearly the entire world searches the Internet.

Adam Epstein—the president and co-CEO of adMarketplace, which bills itself as “the largest consumer search technology company outside of Google and Bing”—told Ars that innovations in search could result in a broader landscape of more dynamic search experiences that draw from sources beyond Google and allow searchers to skip Google’s SERPs entirely. If that happens, the coming years could make Google’s ubiquitous search experience today a distant memory.

“By the end of this decade, going to a search engine results page will seem quaint,” Epstein predicted. “The court’s decision sets the stage for a remedy that will dramatically improve the search experience for everyone connected to the web. The era of innovation in search is just around the corner.”

The DOJ has not meaningfully discussed potential remedies it will seek, but Jonathan Kanter, assistant attorney general of the Justice Department’s antitrust division, celebrated the ruling.

“This landmark decision holds Google accountable,” Kanter said. “It paves the path for innovation for generations to come and protects access to information for all Americans.”

All the possible ways to destroy Google’s monopoly in search Read More »

google-antitrust-verdict-leaves-apple-with-“inconvenient-alternatives”

Google antitrust verdict leaves Apple with “inconvenient alternatives”

trustbusting —

A reliable source of billions of dollars in income is at risk for the iPhone maker.

A Google

Benj Edwards

The landmark antitrust ruling against Google on Monday is shaking up one of the longest-standing partnerships in tech.

At the heart of the case are billions of dollars’ worth of exclusive agreements Google has inked over the years to become the default search engine on browsers and devices across the world. No company benefited more than fellow Big Tech giant Apple—which US District Judge Amit Mehta called a “crucial partner” to Google.

During a weekslong trial, Apple executives showed up to explain and defend the partnership. Under a deal that first took shape in 2002, Google paid a cut of search advertising revenue to Apple to direct its users to Google Search as default, with payments reaching $20 billion for 2022, according to the court’s findings. In exchange, Google got access to Apple’s valuable user base—more than half of all search queries in the US currently flow through Apple devices.

Since Monday’s ruling, Apple has been quiet. But it is likely to be deeply involved in the next phase of the case, which will address the proposed fix to Google’s legal breaches. Remedies in the case could be targeted or wide-ranging. The Department of Justice, which brought the case, has not said what it will seek.

“The most profound impact of the judgment is liable to be felt by Apple,” said Eric Seufert, an independent analyst.

JPMorgan analysts wrote that the ruling left Apple with a range of “inconvenient alternatives,” including the possibility of a new revenue-sharing agreement with Google that does not grant it exclusive rights as the default search engine, thereby reducing its value.

Reaching revenue-sharing deals with alternative search engines like Microsoft’s Bing, they wrote, would “offer lower economic benefits for Apple, given Google’s superior advertising monetisation.”

Mehta noted in his ruling that the idea of replacing the Google agreement with one involving Microsoft and Bing had come up previously. Eddy Cue, Apple’s senior vice-president of services, “concluded that a Microsoft-Apple deal would only make sense if Apple ‘view[ed] Google as somebody [they] don’t want to be in business with and therefore are willing to jeopardize revenue to get out. Otherwise it [was a] no brainer to stay with Google as it is as close to a sure thing as can be,’” Mehta wrote.

Apple could build its own search engine. It has not yet done so, and the judge in the case stopped short of agreeing with the DoJ that the Google deal amounted to a “pay-off” to Apple to keep it out of the search engine market. An internal Apple study in 2018, cited in the judge’s opinion, found that even if it did so and maintained 80 percent of queries, it would still lose $12 billion in revenue in the first five years after separating from Google.

Mehta cited an email from John Giannandrea, a former Google executive who now works for Apple, saying, “There is considerable risk that [Apple] could end up with an unprofitable search engine that [is] also not better for users.”

Google has vowed to appeal against the ruling. Nicholas Rodelli, an analyst at CRFA Research, said it was a “long shot,” given the “meticulous” ruling.

Rodelli said he believed the judge “isn’t likely to issue a game-changing injunction,” such as a full ban on revenue-sharing with Apple. Depending on the remedy the judge decides for Google’s antitrust violations, Seufert said Apple could “either be forced to accept a much less lucrative arrangement with Microsoft [over Bing] or may be prevented from selling search defaults at all.”

“It’s certainly going to adjust the relationship between Google and Apple,” said Bill Kovacic, a former Federal Trade Commission chair and professor of competition law and policy at George Washington University Law School.

Mozilla’s funding may be at risk

Apple is not the only company potentially affected by Monday’s ruling. According to the court, Google’s 2021 payment to Mozilla for the default position on its browser was more than $400 million, about 80 percent of Mozilla’s operating budget. A spokesperson for Mozilla said it was “closely reviewing” the decision and “how we can positively influence the next steps.”

Meanwhile, the search market is undergoing a transformation, as companies such as Google and Microsoft explore how generative AI chatbots can transform traditional search features.

Apple’s partnership with OpenAI, announced in June, will allow users to direct their queries to its chatbot ChatGPT. A smarter Siri voice assistant powered by Apple’s own proprietary AI models will also create a new outlet for user queries that might otherwise go to Google. Apple’s models are trained using Applebot, a web crawler that, much like the technology behind a search engine, compiles public information from across the Internet.

Traditional search is showing no signs of slowing. Research from Emarketer finds that, in the US alone, spend on search advertising will grow at an average of about 10 percent each year, hitting $184 billion in 2028. Google, the dominant player by a long shot, captures about half of that spend. Apple’s current deal with Google would have allowed it to unilaterally extend the partnership into 2028.

The Cupertino, California-based iPhone maker has its own antitrust battle to wage. The DoJ’s antitrust division, led by Jonathan Kanter, filed a sweeping lawsuit against Apple in March, making it the latest Big Tech giant to be targeted by the Biden administration’s enforcers.

The legal troubles reflect an ongoing decline in Apple’s relationship with policymakers in Washington, despite an effort by chief executive Tim Cook to step up the company’s lobbying of the Biden White House, according to research by the Tech Transparency Project. TTP found that Apple spent $9.9 million on lobbying the federal government in 2023—its highest in 25 years, though still much lower than the likes of Google, Amazon, and Meta.

© 2024 The Financial Times Ltd. All rights reserved. Not to be redistributed, copied, or modified in any way.

Google antitrust verdict leaves Apple with “inconvenient alternatives” Read More »

google-loses-doj’s-big-monopoly-trial-over-search-business

Google loses DOJ’s big monopoly trial over search business

Huge loss for Google —

Google’s exclusive deals maintained monopolies in two markets, judge ruled.

Google loses DOJ’s big monopoly trial over search business

Google just lost a massive antitrust trial over its sprawling search business, as US district judge Amit Mehta released his ruling, showing that he sided with the US Department of Justice in the case that could disrupt how billions of people search the web.

“Google is a monopolist, and it has acted as one to maintain its monopoly,” Mehta wrote in his opinion. “It has violated Section 2 of the Sherman Act.”

The verdict will likely come as a shock to Google, which had long argued that punishing Google for being the best in search would be “unprecedented” and frequently pointed to the DOJ’s lack of direct evidence. However, Mehta found the limited direct evidence compelling, especially “Google’s admission that it does not ‘consider whether users will go to other specific search providers (general or otherwise) if it introduces a change to its Search product.'”

“Google’s indifference is unsurprising,” Mehta wrote. “In 2020, Google conducted a quality degradation study, which showed that it would not lose search revenue if were to significantly reduce the quality of its search product. Just as the power to raise price ‘when it is desired to do so’ is proof of monopoly power, so too is the ability to degrade product quality without concern of losing consumers.”

He also wrote that the DOJ’s indirect evidence “easily establishes Google’s monopoly power in search” and concluded that “the fact that Google makes product changes without concern that its users might go elsewhere is something only a firm with monopoly power could do.”

Google didn’t lose every battle in this big fight with the DOJ. Mehta ruled that Google did not have monopoly power in search advertising, agreed that there was no market for general search advertising, and declined to sanction Google for allegedly destroying evidence by “failing to preserve its employees’ chat messages.”

Google’s president of global affairs, Kent Walker, provided a statement to Ars, confirming that Google plans to appeal.

“This decision recognizes that Google offers the best search engine, but concludes that we shouldn’t be allowed to make it easily available,” Walker said. “We appreciate the Court’s finding that Google is ‘the industry’s highest quality search engine, which has earned Google the trust of hundreds of millions of daily users,’ that Google ‘has long been the best search engine, particularly on mobile devices,’ ‘has continued to innovate in search,’ and that ‘Apple and Mozilla occasionally assess Google’s search quality relative to its rivals and find Google’s to be superior.’ Given this, and that people are increasingly looking for information in more and more ways, we plan to appeal. As this process continues, we will remain focused on making products that people find helpful and easy to use.”

Google monopolizes two markets, judge ruled

Mehta ruled that Google spending billions on exclusive distribution agreements with companies like Apple helped the tech giant maintain monopolies in two markets, general search services and general text advertising.

The US government had argued that Google used these exclusive deals to block out competitors like Bing or DuckDuckGo, “by ensuring that all of Android and Apple and mobile users are offered Google, either as the default general search engine or the only general search engine, Google’s deals with Android and Apple clearly have a significant effect in preserving its monopoly.” The DOJ successfully argued that blocks rivals from reaching the “critical level necessary” to “pose a real threat to Google’s monopoly.”

Mehta noted that Google’s dominance had “gone unchallenged for well over a decade,” partly due to a “largely unseen advantage over its rivals: default distribution.” He found that Google’s exclusive distribution deals foreclosed a “substantial share” of the markets and allowed Google to earn more revenues. Google then shared spiking revenues with device and browser developers—spending up to $26 billion in 2021 alone for exclusive deals, the trial revealed.

Google did all this, Mehta said, to ensure that “most devices in the United States come preloaded exclusively with Google” and to force “Google’s rivals to find other ways to reach users.” The DOJ successfully argued that this posed “significant barriers that protect Google’s market dominance in general search,” with rivals having to overcome “high capital costs—”to the tune of billions of dollars,” Mehta wrote—”Google’s control of key distribution channels, brand recognition, and scale.”

Barriers to entry in general text advertising are similarly “high,” Mehta said, with new entrants facing “the same major obstacles as would the developer of a new” search engine.

One of the most scrutinized exclusive deals was between Google and Apple, which was estimated at $20 billion in 2022. “This is nearly double the payment made in 2020,” Mehta noted, suggesting that Google increasingly valued the deal locking its search engine as the default in Safari as a way to shore up its search dominance.

“Google has long recognized that, if Apple were to develop and deploy its own search engine as the default” search tool “in Safari, it would come at great cost to Google,” Mehta wrote. Without the deal, Google “would lose around 65 percent of its revenue, even assuming that it could retain some users without the Safari default” placement. But “Apple has decided not to enter general search,” Mehta said, likely because it “would forego significant revenues” and potentially face user backlash if it stopped partnering with Google. Similarly high revenue loss would occur if “Google were to lose the Android defaults,” Mehta said.

None of the pro-competitive benefits that Google claimed justified the exclusive deals persuaded Mehta, who ruled that “importantly,” Google “exercised its monopoly power by charging supracompetitive prices for general search text ads”—and thus earned “monopoly profits.”

“That Google makes changes to its text ads auctions without considering its rivals’ prices is something that only a firm with monopoly power is able to do,” Mehta wrote. And “Google in fact has profitably raised prices substantially above the competitive level. That makes ‘the existence of monopoly power” “clear.”

Ultimately, Mehta ruled that “Google has no true competitor” in general search and without any “genuine” competition, “over the last decade, Google’s grip on the market has only grown stronger.” Further, he found that “Google understands there is no genuine competition for the defaults because it knows that its partners cannot afford to go elsewhere,” disagreeing with Google’s arguments that the default deals were not exclusive.

“The key question then is this: Do Google’s exclusive distribution contracts reasonably appear capable of significantly contributing to maintaining Google’s monopoly power in the general search services market?” Mehta wrote. “The answer is ‘yes.'”

This is a developing story and is being updated.

Google loses DOJ’s big monopoly trial over search business Read More »

chrome’s-manifest-v3,-and-its-changes-for-ad-blocking,-are-coming-real-soon

Chrome’s Manifest V3, and its changes for ad blocking, are coming real soon

Chrome Manifest V3 —

Chrome is warning users that their extension makers need to update soon.

Chrome logo, squared off in the style of a popular ad-blocking logo

Ron Amadeo

Google Chrome’s long, long project to implement a new browser extension platform is seemingly going to happen, for real, after six years of cautious movement.

One of the first ways people are seeing this is if they use uBlock Origin, a popular ad-blocking extension, as noted by Bleeping Computer. Recently, Chrome users have seen warnings pop up that “This extension may soon no longer be supported,” with links asking the user to “Remove or replace it with similar extensions” from Chrome’s Web Store. You might see a similar warning on some extensions if you head to Chrome’s Extensions page (chrome://extensions).

What’s happening is Chrome preparing to make Manifest V3 required for extensions that want to run on its platform. First announced in 2018, the last word on Manifest V3 was that V2 extensions would start being nudged out in early June on the Beta, Dev, and Canary update channels. Users will be able to manually re-enable V2 extensions “for a short time,” Google has said, “but over time, this toggle will go away as well.” The shift for enterprise Chrome deployments is expected to be put off until June 2025.

Google has said that its new extension platform was built for “improving the security, privacy, performance, and trustworthiness of the extension ecosystem.” The Electronic Frontier Foundation (EFF) disagrees most strongly with the security aspect, and Firefox-maker Mozilla, while intending to support V3 extensions for cross-browser compatibility, has no plans to cut off support for V2 extensions, signaling that it doesn’t see the big improvement.

Perhaps the biggest point of friction is with ad blockers. Google has said it “isn’t killing ad blockers” but “making them safer,” in an explanatory blog post. Google noted in November 2023 that Manifest V3 allowed for a greater number, and more dynamic updating, of content-blocking rules in extensions, specifically ad blockers.

But one of the biggest changes is in disallowing “remotely hosted code,” which includes the filtering lists that ad blockers keep regularly updated. Ad blockers that want to update their filtering lists, perhaps in response to pivots by platforms like Google’s YouTube and ad servers, will have to do so through the Chrome Web Store’s review process. Ad-blocking coders see it as an intentional gatekeeping and slowing.

Google said before the initial May push toward V3 that 85 percent of actively maintained extensions in its store had Manifest V3 versions ready. Raymond Hill wrote on uBlock Origin’s GitHub page Friday that there will not be a full version of uBlock Origin that works with Manifest V3, but instead a “Lite” version that is “a pared-down version of uBO with a best effort at converting filter lists used by uBO into a Manifest V3-compliant approach.”

Chrome’s Manifest V3, and its changes for ad blocking, are coming real soon Read More »

google-pulls-its-terrible-pro-ai-“dear-sydney”-ad-after-backlash

Google pulls its terrible pro-AI “Dear Sydney” ad after backlash

Gemini, write me a fan letter! —

Taking the “human” out of “human communication.”

A picture of the Gemini prompt box from the

Enlarge / The Gemini prompt box in the “Dear Sydney” ad.

Google

Have you seen Google’s “Dear Sydney” ad? The one where a young girl wants to write a fan letter to Olympic hurdler Sydney McLaughlin-Levrone? To which the girl’s dad responds that he is “pretty good with words but this has to be just right”? And so, to be just right, he suggests that the daughter get Google’s Gemini AI to write a first draft of the letter?

If you’re watching the Olympics, you have undoubtedly seen it—because the ad has been everywhere. Until today. After a string of negative commentary about the ad’s dystopian implications, Google has pulled the “Dear Sydney” ad from TV. In a statement to The Hollywood Reporter, the company said, “While the ad tested well before airing, given the feedback, we have decided to phase the ad out of our Olympics rotation.”

The backlash was similar to that against Apple’s recent ad in which an enormous hydraulic press crushed TVs, musical instruments, record players, paint cans, sculptures, and even emoji into… the newest model of the iPad. Apple apparently wanted to show just how much creative and entertainment potential the iPad held; critics read the ad as a warning image about the destruction of human creativity in a technological age. Apple apologized soon after.

Now Google has stepped on the same land mine. Not only is AI coming for human creativity, the “Dear Sydney” ad suggests—but it won’t even leave space for the charming imperfections of a child’s fan letter to an athlete. Instead, AI will provide the template, just as it will likely provide the template for the athlete’s response, leading to a nightmare scenario in which huge swathes of human communication have the “human” part stripped right out.

“Very bad”

The generally hostile tone of the commentary to the new ad was captured by Alexandra Petri’s Washington Post column on the ad, which Petri labeled “very bad.”

This ad makes me want to throw a sledgehammer into the television every time I see it. Given the choice between watching this ad and watching the ad about how I need to be giving money NOW to make certain that dogs do not perish in the snow, I would have to think long and hard. It’s one of those ads that makes you think, perhaps evolution was a mistake and our ancestor should never have left the sea. This could be slight hyperbole but only slight!

If you haven’t seen this ad, you are leading a blessed existence and I wish to trade places with you.

A TechCrunch piece said that it was “hard to think of anything that communicates heartfelt inspiration less than instructing an AI to tell someone how inspiring they are.”

Shelly Palmer, a Syracuse University professor and marketing consultant, wrote that the ad’s basic mistake was overestimating “AI’s ability to understand and convey the nuances of human emotions and thoughts.” Palmer would rather have a “heartfelt message over a grammatically correct, AI-generated message any day,” he said. He then added:

I received just such a heartfelt message from a reader years ago. It was a single line email about a blog post I had just written: “Shelly, you’re to [sic] stupid to own a smart phone.” I love this painfully ironic email so much, I have it framed on the wall in my office. It was honest, direct, and probably accurate.

But his conclusion was far more serious. “I flatly reject the future that Google is advertising,” Palmer wrote. “I want to live in a culturally diverse world where billions of individuals use AI to amplify their human skills, not in a world where we are used by AI pretending to be human.”

Things got saltier from there. NPR host Linda Holmes wrote on social media:

This commercial showing somebody having a child use AI to write a fan letter to her hero SUCKS. Obviously there are special circumstances and people who need help, but as a general “look how cool, she didn’t even have to write anything herself!” story, it SUCKS. Who wants an AI-written fan letter?? I promise you, if they’re able, the words your kid can put together will be more meaningful than anything a prompt can spit out. And finally: A fan letter is a great way for a kid to learn to write! If you encourage kids to run to AI to spit out words because their writing isn’t great yet, how are they supposed to learn? Sit down with your kid and write the letter with them! I’m just so grossed out by the entire thing.

The Atlantic was more succinct with its headline: “Google Wins the Gold Medal for Worst Olympic Ad.”

All of this largely tracks with our own take on the ad, which Ars Technica’s Kyle Orland called a “grim” vision of the future. “I want AI-powered tools to automate the most boring, mundane tasks in my life, giving me more time to spend on creative, life-affirming moments with my family,” he wrote. “Google’s ad seems to imply that these life-affirming moments are also something to be avoided—or at least made pleasingly more efficient—through the use of AI.”

Getting people excited about their own obsolescence and addiction is a tough sell, so I don’t envy the marketers who have to hawk Big Tech’s biggest products in a climate of suspicion and hostility toward everything from AI to screen time to social media to data collection. I’m sure the marketers will find a way—but clearly “Dear Sydney” isn’t it.

Google pulls its terrible pro-AI “Dear Sydney” ad after backlash Read More »

google-won’t-downrank-top-deepfake-porn-sites-unless-victims-mass-report

Google won’t downrank top deepfake porn sites unless victims mass report

Google won’t downrank top deepfake porn sites unless victims mass report

Today, Google announced new measures to combat the rapidly increasing spread of AI-generated non-consensual explicit deepfakes in its search results.

Because of “a concerning increase in generated images and videos that portray people in sexually explicit contexts, distributed on the web without their consent,” Google said that it consulted with “experts and victim-survivors” to make some “significant updates” to its widely used search engine to “further protect people.”

Specifically, Google made it easier for targets of fake explicit images—which experts have said are overwhelmingly women—to report and remove deepfakes that surface in search results. Additionally, Google took steps to downrank explicit deepfakes “to keep this type of content from appearing high up in Search results,” the world’s leading search engine said.

Victims of deepfake pornography have previously criticized Google for not being more proactive in its fight against deepfakes in search results. Surfacing images and reporting each one is a “time- and energy-draining process” and “constant battle,” Kaitlyn Siragusa, a Twitch gamer with an explicit OnlyFans frequently targeted by deepfakes, told Bloomberg last year.

In response, Google has worked to “make the process easier,” partly by “helping people address this issue at scale.” Now, when a victim submits a removal request, “Google’s systems will also aim to filter all explicit results on similar searches about them,” Google’s blog said. And once a deepfake is “successfully removed,” Google “will scan for—and remove—any duplicates of that image that we find,” the blog said.

Google’s efforts to downrank harmful fake content have also expanded, the tech giant said. To help individuals targeted by deepfakes, Google will now “lower explicit fake content for” searches that include people’s names. According to Google, this step alone has “reduced exposure to explicit image results on these types of queries by over 70 percent.”

However, Google still seems resistant to downranking general searches that might lead people to harmful content. A quick Google search confirms that general searches with keywords like “celebrity nude deepfake” point searchers to popular destinations where they can search for non-consensual intimate images of celebrities or request images of less famous people.

For victims, the bottom line is that problematic links will still appear in Google’s search results for anyone willing to keep scrolling or anyone intentionally searching for “deepfakes.” The only step Google has taken recently to downrank top deepfake sites like Fan-Topia or MrDeepFakes is a promise to demote “sites that have received a high volume of removals for fake explicit imagery.”

It’s currently unclear what Google considers a “high volume,” and Google declined Ars’ request to comment on whether these sites would be downranked eventually. Instead, a Google spokesperson told Ars that “if we receive a high volume of successful removal sites from a specific website under this policy, we will use that as a ranking signal and demote the site in question for queries where the site might surface.”

Currently, Google’s spokesperson said, Google is focused on downranking “queries that include the names of individuals,” which “have the highest potential for individual harm.” But more queries will be downranked in the coming months, Google’s spokesperson said, and Google continues to tackle the “technical challenge for search engines” of differentiating between “explicit content that’s real and consensual (like an actor’s nude scenes)” and “explicit fake content (like deepfakes featuring said actor),” Google’s blog said.

“This is an ongoing effort, and we have additional improvements coming over the next few months to address a broader range of queries,” Google’s spokesperson told Ars.

Deepfake trauma “never ends”

In its blog, Google said that “these efforts are designed to give people added peace of mind, especially if they’re concerned about similar content about them popping up in the future.”

But many deepfake victims have claimed that dedicating hours or even months to removing harmful content doesn’t provide any hope that the images won’t resurface. Most recently, one deepfake victim, Sabrina Javellana, told The New York Times that even after her home state Florida passed a law against deepfakes, that didn’t stop the fake images from spreading online.

She’s given up on trying to get the images removed anywhere, telling The Times, “It just never ends. I just have to accept it.”

According to US Representative Joseph Morelle (D-NY), it will take a federal law against deepfakes to deter more bad actors from harassing and terrorizing women with deepfake porn. He’s introduced one such law, the Preventing Deepfakes of Intimate Images Act, which would criminalize creating deepfakes. It currently has 59 sponsors in the House and bipartisan support in the Senate, Morelle said on a panel this week discussing harms of deepfakes, which Ars attended.

Morelle said he’d spoken to victims of deepfakes, including teenagers, and decided that “a national ban and a national set of both criminal and civil remedies makes the most sense” to combat the problem with “urgency.”

“A patchwork of different state and local jurisdictions with different rules” would be “really hard to follow” for both victims and perpetrators trying to understand what’s legal, Morelle said, whereas federal laws that impose a liability and criminal penalty would likely have “the greatest impact.”

Victims, Morelle said, every day suffer from mental, physical, emotional, and financial harms, and as a co-panelist, Andrea Powell, pointed out, there is no healing because there is currently no justice for survivors during a period of “prolific and catastrophic increase in this abuse,” Powell warned.

Google won’t downrank top deepfake porn sites unless victims mass report Read More »

outsourcing-emotion:-the-horror-of-google’s-“dear-sydney”-ai-ad

Outsourcing emotion: The horror of Google’s “Dear Sydney” AI ad

Here's an idea: Don't be a deadbeat and do it yourself!

Enlarge / Here’s an idea: Don’t be a deadbeat and do it yourself!

If you’ve watched any Olympics coverage this week, you’ve likely been confronted with an ad for Google’s Gemini AI called “Dear Sydney.” In it, a proud father seeks help writing a letter on behalf of his daughter, who is an aspiring runner and superfan of world-record-holding hurdler Sydney McLaughlin-Levrone.

“I’m pretty good with words, but this has to be just right,” the father intones before asking Gemini to “Help my daughter write a letter telling Sydney how inspiring she is…” Gemini dutifully responds with a draft letter in which the LLM tells the runner, on behalf of the daughter, that she wants to be “just like you.”

Every time I see this ad, it puts me on edge in a way I’ve had trouble putting into words (though Gemini itself has some helpful thoughts). As someone who writes words for a living, the idea of outsourcing a writing task to a machine brings up some vocational anxiety. And the idea of someone who’s “pretty good with words” doubting his abilities when the writing “has to be just right” sets off alarm bells regarding the superhuman framing of AI capabilities.

But I think the most offensive thing about the ad is what it implies about the kinds of human tasks Google sees AI replacing. Rather than using LLMs to automate tedious busywork or difficult research questions, “Dear Sydney” presents a world where Gemini can help us offload a heartwarming shared moment of connection with our children.

The “Dear Sydney” ad.

It’s a distressing answer to what’s still an incredibly common question in the AI space: What do you actually use these things for?

Yes, I can help

Marketers have a difficult task when selling the public on their shiny new AI tools. An effective ad for an LLM has to make it seem like a superhuman do-anything machine but also an approachable, friendly helper. An LLM has to be shown as good enough to reliably do things you can’t (or don’t want to) do yourself, but not so good that it will totally replace you.

Microsoft’s 2024 Super Bowl ad for Copilot is a good example of an attempt to thread this needle, featuring a handful of examples of people struggling to follow their dreams in the face of unseen doubters. “Can you help me?” those dreamers ask Copilot with various prompts. “Yes, I can help” is the message Microsoft delivers back, whether through storyboard images, an impromptu organic chemistry quiz, or “code for a 3D open world game.”

Microsoft’s Copilot marketing sells it as a helper for achieving your dreams.

The “Dear Sydney” ad tries to fit itself into this same box, technically. The prompt in the ad starts with “Help my daughter…” and the tagline at the end offers “A little help from Gemini.” If you look closely near the end, you’ll also see Gemini’s response starts with “Here’s a draft to get you started.” And to be clear, there’s nothing inherently wrong with using an LLM as a writing assistant in this way, especially if you have a disability or are writing in a non-native language.

But the subtle shift from Microsoft’s “Help me” to Google’s “Help my daughter” changes the tone of things. Inserting Gemini into a child’s heartfelt request for parental help makes it seem like the parent in question is offloading their responsibilities to a computer in the coldest, most sterile way possible. More than that, it comes across as an attempt to avoid an opportunity to bond with a child over a shared interest in a creative way.

It’s one thing to use AI to help you with the most tedious parts of your job, as people do in recent ads for Salesforce’s Einstein AI. It’s another to tell your daughter to go ask the computer for help pouring their heart out to their idol.

Outsourcing emotion: The horror of Google’s “Dear Sydney” AI ad Read More »

loss-of-popular-2fa-tool-puts-security-minded-grapheneos-in-a-paradox

Loss of popular 2FA tool puts security-minded GrapheneOS in a paradox

Just a bit too custom for their taste —

Losing access to Authy leads to another reckoning with Google’s security model.

Scientist looking at a molecular model of graphene in a laboratory

Enlarge / Graphene is a remarkable allotrope, deserving of further study. GrapheneOS is a remarkable ROM, one that Google does not quite know how to accommodate, due to its “tiny, tiny” user numbers compared to mainstream Android.

“If it’s not an official OS, we have to assume it’s bad.”

That’s how Shawn Wilden, the tech lead for hardware-backed security in Android, described the current reality of custom Android-based operating systems in response to a real security conundrum. GrapheneOS users discovered recently that Authy, a popular (and generally well-regarded) two-factor authentication manager, will not work on their phones—phones running an OS intended to be more secure and hardened than any standard Android phone.

“We don’t want to punish users of alternative OSes, but there’s really no other option at the moment,” Wilden added before his blunt conclusion. “Play Integrity has absolutely no way to guess whether a given custom OS completely subverts the Android security model.”

Play Integrity, formerly SafetyNet Attestation, essentially allows apps to verify whether an Android device has provided permissions beyond Google’s intended models or has been rooted. Root access is not appealing to the makers of some apps involving banking, payments, competitive games, and copyrighted media.]

There are many reasons beyond cheating and skulduggery that someone might root or modify their Android device. But to prove itself secure, an Android device must contact Google’s servers through an API in Google Play Services and then have its bootloader, ROM signature, and kernel verified. GrapheneOS, like most custom Android ROMs, does not contain a Google Play Services package by default but will let users install a sandboxed version of Play Services if they wish.

Wilden offered some hope for a future in which ROMs could vouch for their non-criminal nature to Google, noting “some discussions with makers of high-quality ROMs” about passing the Compatibility Test Suite, then “establishing some kind of relationship we can use to trust them.” But it’s “a lot of work on both sides, including by lawyers,” Wilden notes. And while his team is happy to help, higher-level support is tough because “modders are such a tiny, tiny fraction of the user base.”

The official GrapheneOS X account was less hopeful. It noted that another custom ROM, LineageOS, disabled verified boot at installation, and “rolls back security in a lot of other ways,” contributing to “a misconception that every alternate OS rolls back security and isn’t production quality.” A typical LineageOS installation, like most custom ROMs, does disable verified boot, though it can be re-enabled, except it’s risky and complicated. GrapheneOS has a page on its site regarding its stance on, and criticisms of, Google’s attestation model for Android.

Ars has reached out to Google, GrapheneOS, and Authy (via owner Twilio) for comment. At the moment, it doesn’t seem like there’s a clear path forward for any party unless one of them is willing to majorly rework what they consider proper security.

Loss of popular 2FA tool puts security-minded GrapheneOS in a paradox Read More »

google-claims-math-breakthrough-with-proof-solving-ai-models

Google claims math breakthrough with proof-solving AI models

slow and steady —

AlphaProof and AlphaGeometry 2 solve problems, with caveats on time and human assistance.

An illustration provided by Google.

Enlarge / An illustration provided by Google.

On Thursday, Google DeepMind announced that AI systems called AlphaProof and AlphaGeometry 2 reportedly solved four out of six problems from this year’s International Mathematical Olympiad (IMO), achieving a score equivalent to a silver medal. The tech giant claims this marks the first time an AI has reached this level of performance in the prestigious math competition—but as usual in AI, the claims aren’t as clear-cut as they seem.

Google says AlphaProof uses reinforcement learning to prove mathematical statements in the formal language called Lean. The system trains itself by generating and verifying millions of proofs, progressively tackling more difficult problems. Meanwhile, AlphaGeometry 2 is described as an upgraded version of Google’s previous geometry-solving AI modeI, now powered by a Gemini-based language model trained on significantly more data.

According to Google, prominent mathematicians Sir Timothy Gowers and Dr. Joseph Myers scored the AI model’s solutions using official IMO rules. The company reports its combined system earned 28 out of 42 possible points, just shy of the 29-point gold medal threshold. This included a perfect score on the competition’s hardest problem, which Google claims only five human contestants solved this year.

A math contest unlike any other

The IMO, held annually since 1959, pits elite pre-college mathematicians against exceptionally difficult problems in algebra, combinatorics, geometry, and number theory. Performance on IMO problems has become a recognized benchmark for assessing an AI system’s mathematical reasoning capabilities.

Google states that AlphaProof solved two algebra problems and one number theory problem, while AlphaGeometry 2 tackled the geometry question. The AI model reportedly failed to solve the two combinatorics problems. The company claims its systems solved one problem within minutes, while others took up to three days.

Google says it first translated the IMO problems into formal mathematical language for its AI model to process. This step differs from the official competition, where human contestants work directly with the problem statements during two 4.5-hour sessions.

Google reports that before this year’s competition, AlphaGeometry 2 could solve 83 percent of historical IMO geometry problems from the past 25 years, up from its predecessor’s 53 percent success rate. The company claims the new system solved this year’s geometry problem in 19 seconds after receiving the formalized version.

Limitations

Despite Google’s claims, Sir Timothy Gowers offered a more nuanced perspective on the Google DeepMind models in a thread posted on X. While acknowledging the achievement as “well beyond what automatic theorem provers could do before,” Gowers pointed out several key qualifications.

“The main qualification is that the program needed a lot longer than the human competitors—for some of the problems over 60 hours—and of course much faster processing speed than the poor old human brain,” Gowers wrote. “If the human competitors had been allowed that sort of time per problem they would undoubtedly have scored higher.”

Gowers also noted that humans manually translated the problems into the formal language Lean before the AI model began its work. He emphasized that while the AI performed the core mathematical reasoning, this “autoformalization” step was done by humans.

Regarding the broader implications for mathematical research, Gowers expressed uncertainty. “Are we close to the point where mathematicians are redundant? It’s hard to say. I would guess that we’re still a breakthrough or two short of that,” he wrote. He suggested that the system’s long processing times indicate it hasn’t “solved mathematics” but acknowledged that “there is clearly something interesting going on when it operates.”

Even with these limitations, Gowers speculated that such AI systems could become valuable research tools. “So we might be close to having a program that would enable mathematicians to get answers to a wide range of questions, provided those questions weren’t too difficult—the kind of thing one can do in a couple of hours. That would be massively useful as a research tool, even if it wasn’t itself capable of solving open problems.”

Google claims math breakthrough with proof-solving AI models Read More »

chrome-will-now-prompt-some-users-to-send-passwords-for-suspicious-files

Chrome will now prompt some users to send passwords for suspicious files

SAFE BROWSING —

Google says passwords and files will be deleted shortly after they are deep-scanned.

Chrome will now prompt some users to send passwords for suspicious files

Google is redesigning Chrome malware detections to include password-protected executable files that users can upload for deep scanning, a change the browser maker says will allow it to detect more malicious threats.

Google has long allowed users to switch on the Enhanced Mode of its Safe Browsing, a Chrome feature that warns users when they’re downloading a file that’s believed to be unsafe, either because of suspicious characteristics or because it’s in a list of known malware. With Enhanced Mode turned on, Google will prompt users to upload suspicious files that aren’t allowed or blocked by its detection engine. Under the new changes, Google will prompt these users to provide any password needed to open the file.

Beware of password-protected archives

In a post published Wednesday, Jasika Bawa, Lily Chen, and Daniel Rubery of the Chrome Security team wrote:

Not all deep scans can be conducted automatically. A current trend in cookie theft malware distribution is packaging malicious software in an encrypted archive—a .zip, .7z, or .rar file, protected by a password—which hides file contents from Safe Browsing and other antivirus detection scans. In order to combat this evasion technique, we have introduced two protection mechanisms depending on the mode of Safe Browsing selected by the user in Chrome.

Attackers often make the passwords to encrypted archives available in places like the page from which the file was downloaded, or in the download file name. For Enhanced Protection users, downloads of suspicious encrypted archives will now prompt the user to enter the file’s password and send it along with the file to Safe Browsing so that the file can be opened and a deep scan may be performed. Uploaded files and file passwords are deleted a short time after they’re scanned, and all collected data is only used by Safe Browsing to provide better download protections.

Enter a file password to send an encrypted file for a malware scan

Enlarge / Enter a file password to send an encrypted file for a malware scan

Google

For those who use Standard Protection mode which is the default in Chrome, we still wanted to be able to provide some level of protection. In Standard Protection mode, downloading a suspicious encrypted archive will also trigger a prompt to enter the file’s password, but in this case, both the file and the password stay on the local device and only the metadata of the archive contents are checked with Safe Browsing. As such, in this mode, users are still protected as long as Safe Browsing had previously seen and categorized the malware.

Sending Google an executable casually downloaded from a site advertising a screensaver or media player is likely to generate little if any hesitancy. For more sensitive files such as a password-protected work archive, however, there is likely to be more pushback. Despite the assurances the file and password will be deleted promptly, things sometimes go wrong and aren’t discovered for months or years, if at all. People using Chrome with Enhanced Mode turned on should exercise caution.

A second change Google is making to Safe Browsing is a two-tiered notification system when users are downloading files. They are:

  1. Suspicious files, meaning those Google’s file-vetting engine have given a lower-confidence verdict, with unknown risk of user harm
  2. Dangerous files, or those with a high confidence verdict that they pose a high risk of user harm

The new tiers are highlighted by iconography, color, and text in an attempt to make it easier for users to easily distinguish between the differing levels of risk. “Overall, these improvements in clarity and consistency have resulted in significant changes in user behavior, including fewer warnings bypassed, warnings heeded more quickly, and all in all, better protection from malicious downloads,” the Google authors wrote.

Previously, Safe Browsing notifications looked like this:

Differentiation between suspicious and dangerous warnings.

Enlarge / Differentiation between suspicious and dangerous warnings.

Google

Over the past year, Chrome hasn’t budged on its continued support of third-party cookies, a decision that allows companies large and small to track users of that browser as they navigate from website to website to website. Google’s alternative to tracking cookies, known as the Privacy Sandbox, has also received low marks from privacy advocates because it tracks user interests based on their browser usage.

That said, Chrome has long been a leader in introducing protections, such as a security sandbox that cordons off risky code so it can’t mingle with sensitive data and operating system functions. Those who stick with Chrome should at a minimum keep Standard Mode Safe Browsing on. Users with the experience required to judiciously choose which files to send to Google should consider turning on Enhanced Mode.

Chrome will now prompt some users to send passwords for suspicious files Read More »

ars-is-seeking-a-seasoned-senior-reporter-for-all-things-google

Ars is seeking a seasoned senior reporter for all things Google

get your ron on —

Got feelings about the future of AI and/or phone bezel width? Come apply!

A photograph of

Enlarge / If you get hired for this position, you’ll be provided an assistant. It’s this guy. This guy is your assistant. His name is “Googly.”

Google is a company in transformation—but “from what and “to what are not always clear. To catalog and examine Google’s moves in this new era of generative AI, Ars Technica is hiring a Senior Technology Reporter to focus on Google, AI, Android, and search. While attention to so-called “consumer products” will be important, this role will be more focused on Google’s big moves as a technology and infrastructure company, moves often made to counter perceived threats from companies like OpenAI, Microsoft, and Perplexity. Informed skepticism is the rule around here, so we’re looking for someone with the chops to bring a critical eye to some deep technical and business issues.

As this is a senior role owning an important beat, it is not an entry-level position. We’re looking for someone who can primarily self-direct when it comes to their reporting and someone who is comfortable working remotely within a similarly remote team. We’d also like someone who can bring to the table deep and intelligent analyses on broader Google topics while also hitting smaller daily news stories.

This is a full-time union job with benefits.

All candidates:

  • Must have prior professional experience in technology journalism
  • Must be living in and eligible to work in the United States
  • Should expect to travel two to three times per year for major event coverage
  • Must be comfortable with fully remote work

The full job description and official details can all be found at the listing on the Condé Careers site. If this sounds like the job for you, please apply!

Ars is seeking a seasoned senior reporter for all things Google Read More »