The past 12 months have been undeniably busy for those working in AI. There have been more successful product launches than we can count, and even Nobel Prizes. But it hasnt always been smooth sailing.AI is an unpredictable technology, and the increasing availability of generative models has led people to test their limits in new, weird, and sometimes harmful ways. These were some of 2024s biggest AI misfires.AI slop infiltrated almost every corner of the internetGenerative AI makes creating reams of text, images, videos, and other types of material a breeze. Because it takes just a few seconds between entering a prompt for your model of choice to spit out the result, these models have become a quick, easy way to produce content on a massive scale. And 2024 was the year we started calling this (generally poor quality) media what it isAI slop.This low-stakes way of creating AI slop means it can now be found in pretty much every corner of the internet: from the newsletters in your inbox and books sold on Amazon, to ads and articles across the web and shonky pictures on your social media feeds. The more emotionally evocative these pictures are (wounded veterans, crying children, a signal of support in the Israel-Palestine conflict) the more likely they are to be shared, resulting in higher engagement and ad revenue for their savvy creators.AI slop isnt just annoyingits rise poses a genuine problem for the future of the very models that helped to produce it. Because those models are trained on data scraped from the internet, the increasing number of junky websites containing AI garbage means theres a very real danger models output and performance will get steadily worse.AI art is warping our expectations of real events2024 was also the year that the effects of surreal AI images started seeping into our real lives. Willys Chocolate Experience, a wildly unofficial immersive event inspired by Roald Dahls Charlie and the Chocolate Factory, made headlines across the world in February after its fantastical AI-generated marketing materials gave visitors the impression it would be much grander than the sparsely-decorated warehouse its producers created.Similarly, hundreds of people lined the streets of Dublin for a Halloween parade that didnt exist. A Pakistan-based website used AI to create a list of events in the city, which was shared widely across social media ahead of October 31. Although the SEO-baiting site (myspirithalloween.com) has since been taken down, both events illustrate how misplaced public trust in AI-generated material online can come back to haunt us.Grok allows users to create images of pretty much any scenarioThe vast majority of major AI image generators have guardrailsrules that dictate what AI models can and cant doto prevent users from creating violent, explicit, illegal, and other types of harmful content. Sometimes these guardrails are just meant to make sure that no one makes blatant use of others intellectual property. But Grok, an assistant made by Elon Musks AI company, called xAI, ignores almost all of these principles in line with Musks rejection of what he calls woke AI.Whereas other image models will generally refuse to create images of celebrities, copyrighted material, violence, or terrorismunless theyre tricked into ignoring these rulesGrok will happily generate images of Donald Trump firing a bazooka, or Mickey Mouse holding a bomb. While it draws the line at generating nude images, its refusal to play by the rules undermines other companies efforts to steer clear of creating problematic material.Sexually explicit deepfakes of Taylor Swift circulated onlineIn January, non-consensual deepfake nudes of singer Taylor Swift started circulating on social media, including X and Facebook. A Telegram community tricked Microsofts AI image generator Designer into making the explicit images, demonstrating how guardrails can be circumvented even when they are in place.While Microsoft quickly closed the systems loopholes, the incident shone a light on the platforms poor content-moderation policies, after posts containing the images circulated widely and remained live for days. But the most chilling takeaway is how powerless we still are to fight non-consensual deepfake porn. While watermarking and data-poisoning tools can help, theyll need to be adopted much more widely to make a difference.Business chatbots went haywireAs AI becomes more widespread, businesses are racing to adopt generative tools to save time and money, and to maximize efficiency. The problem ischatbots make stuff up and cant be relied upon to always provide you with accurate information.Air Canada found this out the hard way after its chatbot advised a customer to follow a bereavement refund policy that didnt exist. In February, a Canadian small-claims tribunal upheld the customers legal complaint, despite the airlines assertion that the chatbot was a separate legal entity that is responsible for its own actions.In other high-profile examples of how chatbots can do more harm than good, delivery firm DPDs bot cheerfully swore and called itself useless with little prompting, while a different bot set up to provide New Yorkers with accurate information about their citys government ended up dispensing guidance on how to break the law.AI gadgets arent exactly setting the market alightHardware assistants are something the AI industry tried, and failed, to crack in 2024. Humane attempted to sell customers on the promise of the Ai Pin, a wearable lapel computer, but even slashing its price failed to boost weak sales. The Rabbit R1, a ChatGPT-based personal assistant device, suffered a similar fate, following a rash of critical reviews and reports that it was slow and buggy. Both products seemed to be trying to solve a problem that did not actually exist.AI search summaries went awryHave you ever added glue to a pizza, or eaten a small rock? These are just some of the outlandish suggestions that Googles AI Overviews feature gave web users in May after the search giant added generated responses to the top of search results. Because AI systems cant tell the difference between a factually correct news story and a joke post on Reddit, users raced to find the strangest responses AI Overviews could generate.But AI summaries can also have serious consequences. A new iPhone feature that groups app notifications together and creates summaries of their contents, recently generated a false BBC News headline. The summary falsely stated that Luigi Mangione, who has been charged with the murder of healthcare insurance CEO Brian Thompson, had shot himself. The same feature had previously created a headline claiming that Israeli prime minister Benjamin Netanyahu had been arrested, which was also incorrect. These kinds of errors can inadvertently spread misinformation and undermine trust in news organizations.

Former U.S. President Jimmy Carter is being remembered for his impact on homebuilding across the country after his death on December 29th at the age of 100. Both he and his late wife Rosalynn Carter helped increase awareness of Habitat for Humanitys mission after they became involved locally as volunteer builders in their home state of Georgia in the 1980s.The National Association of Homebuilders said: "President Carter was always an advocate for affordable housing. For decades in his post-presidency, Jimmy Carter and his wife, Rosalynn, kept this vital issue front and center [...] With his own hands, President Carter helped to build countless homes for worthy families because he understood that homeownership is vital to the American way of life."We included the nonprofit as part of our guide to the best volunteering opportunities for architects in April of this year.

The dawn of congestion pricing in parts of New York City is upon us for 2025 after a ruling on an ongoinglawsuit cleared the path for $9 tolls to begin in January, reports NYCs local NBC affiliate. The ruling, which is conversely being framed as a victory for petitioners from the State of New Jersey, allows the MTA to move "full speed ahead" on implementation with a remand to answer the courts remaining questions onair quality no later than January 17th. This summer, the issue waspointed to by Kate Wagner as being one of many "quite mundane"cause clbres for architects and planners to rally behind as individual "agents of change."

Another mixed-use development in Shanghai from Foster + Partners, the Sanlin Riverside development, has been added to the pipeline after breaking ground recently. The project, which couples with the Changfeng mixed-use development in the city's Science and Technology Finance hub to become the firms second new Shanghai project weve covered this year, offers users a connection to its waterfront site with a new anchoring landmark building to be used for retail.The firm says each new building in the development will showcase a "modern representation of 20th century Shanghainese architecture" that will "bring a spirit of community to this new part of Shanghai." The country's largest metro has seen an influx of significant new architecture in 2024 with the addition of designs from Safdie Architects, Neri&Hu, andPEI Architects. Snhetta's newGrand Opera Housescheme will provide another headliner soon as we look forward to 2025.

Netflix announced that itsSquid Game: Unleashedmobile game has reachedNo. 1 Top Free Action Game in 57 countriesRead More

Sequential recommendation systems play a key role in creating personalized user experiences across various platforms, but they also face persistent challenges. Traditionally, these systems rely on users interaction histories to predict preferences, often leading to generic recommendations. While integrating auxiliary data such as item descriptions or intent predictions can provide some improvement, these systems struggle to adapt to user preferences in real-time. Additionally, the absence of comprehensive benchmarks for evaluating preference discernment limits the ability to assess their effectiveness in diverse scenarios.To tackle these issues, a team of researchers from Meta AI, ELLIS Unit, LIT AI Lab, Institute for Machine Learning, JKU Linz, Austria, and the University of Wisconsin, Madison, introduces a paradigm called preference discerning, supported by a generative retrieval model named Mender (Multimodal Preference Discerner). This approach explicitly conditions recommendation systems on user preferences expressed in natural language. Leveraging large language models (LLMs), the framework extracts preferences from reviews and item-specific data, transforming them into actionable insights.Mender captures items at two levels of abstraction: semantic IDs and natural language descriptions. This multimodal approach ensures a more nuanced understanding of user preferences. By combining preference approximationderiving preferences from user datawith preference conditioning, Mender allows systems to dynamically adapt to specific user preferences. Additionally, Meta AI has introduced a benchmark that evaluates preference discerning across five dimensions: preference-based recommendation, sentiment following, fine- and coarse-grained steering, and history consolidation, setting a new standard for evaluating personalization.Technical Features and Advantages of MenderMenders design focuses on integrating user preferences with interaction data seamlessly. It uses pre-trained language models to encode preferences and interaction histories in natural language. Its cross-attention mechanisms enable the decoder to predict semantic IDs for recommended items. Mender comes in two variants:MenderTok: Processes preferences and item sequences holistically, supporting fine-tuning.MenderEmb: Precomputes embeddings for efficient training.Key benefits of Mender include:Preference Steering: Tailoring recommendations dynamically based on user-specified preferences.Sentiment Integration: Utilizing user sentiment to enhance accuracy.History Consolidation: Merging new preferences with historical data to refine results.Results and InsightsMeta AIs evaluation of Mender highlights its significant performance improvements on datasets such as Amazon reviews and Steam. For instance:On the Amazon Beauty subset, MenderTok improved Recall@10 by over 45% compared to baseline models.In sentiment following, Mender effectively identified and acted on user sentiments, outperforming other methods by up to 86%.For fine-grained steering, Mender achieved a 70.5% relative improvement, demonstrating its ability to align recommendations with nuanced preferences.ConclusionMeta AIs preference discerning paradigm offers a fresh perspective on sequential recommendation systems, focusing on explicit user preferences articulated in natural language. By integrating LLMs, multimodal representations, and a robust benchmark, this approach improves personalization while providing a framework for future development. With plans to open-source the underlying code and benchmarks, this work has the potential to benefit a broad range of applications, advancing the field of personalized recommendations.Check out the Paper. All credit for this research goes to the researchers of this project. Also,dont forget to follow us onTwitter and join ourTelegram Channel andLinkedIn Group. Dont Forget to join our60k+ ML SubReddit. Asif RazzaqAsif Razzaq is the CEO of Marktechpost Media Inc.. As a visionary entrepreneur and engineer, Asif is committed to harnessing the potential of Artificial Intelligence for social good. His most recent endeavor is the launch of an Artificial Intelligence Media Platform, Marktechpost, which stands out for its in-depth coverage of machine learning and deep learning news that is both technically sound and easily understandable by a wide audience. The platform boasts of over 2 million monthly views, illustrating its popularity among audiences. [Download] Evaluation of Large Language Model Vulnerabilities Report (Promoted)

This article contains spoilers for season 3 of Marvels What If?The third and final season of Marvels What If? is full of new and exciting peeks at worlds that exist outside of the MCUs Sacred Timeline. From old Hollywood glamour to Russian buddy comedy hijinks to old West shootouts, this season certainly has fun playing in the Marvel sandbox. But despite the fact that the majority of these stories are pretty significant deviations from the characters we know in the MCU, this season does do a good job of setting up important parts of upcoming movies Thunderbolts* and Captain America: Brave New World, albeit in a different way than well likely see them on the big screen.ThunderboltsEpisode 4 What IfThe Red Guardian Stopped the Winter Soldier? is a fun buddy comedy that shows us what could have happened if the Red Guardian had joined the Winter Soldier on his mission to retrieve the super soldier serum from Howard and Maria Stark. This accident in the 90s is what kills them in the Sacred Timeline and later kickstarts the events of Captain America: Civil War.In this timeline, however, the Red Guardian keeps the Winter Soldier from killing the Starks. This obviously makes neither the Red Room or Hydra very happy, so Alexei and Bucky end up on the run together for a while. That is, until Bucky is ordered to kill Alexei. Not wanting to kill his new friend, Bucky lets Alexei escape, telling both of their bosses that he died. Even though the two have only known each other for a short time, and we know that Bucky is still very much under Hydras control at this point in the Sacred Timeline, its telling that any version of the Winter Soldier would be willing to risk so much for someone who is basically a stranger.Obviously this story happens at a very different point in these characters lives than Thunderbolts* will, but hopefully this episode still gives us a glimpse of the buddy comedy dynamic we might be able to expect from Alexei and Bucky when we see them in this movie. This episode is also a sign that their first mission together will likely have several hiccups, considering how much the two butt heads in the beginning. However, this episode also shows that these two can also be formidable allies when it really matters.Captain America: Brave New WorldOn the surface, Captain America: Brave New World doesnt look like it would have many ties to the more magical and cosmic sides of the MCU, let alone an episode as glittery and glamorous as What IfAgatha Went to Hollywood? However, this episode of Marvels What If? and Brave New World do have one thing in common the return of the Celestial Tiamut from Eternals.In What If?, Agatha tries to absorb the Celestials power and take it as her own. And she succeeds, becoming an incredibly powerful cosmic being in the process. While we arent quite sure the extent of which the stone corpse of Tiamut will feature in Brave New World, this episode of What If? does show us how much power lies in even a dormant Celestial.Its not likely that anyone will be trying to absorb the Celestials power like Agatha did, at least in Brave New World, but that doesnt mean there isnt going to be a race to claim the new territory and wealth of cosmic resources this stone giant has to offer. In one of the trailers, it looks like Cap (Anthony Mackie) is flying around part of Tiamuts emerging corpse, but weve yet to see just how much the Celestial body is going to factor into the plot and conspiracy at the center of Brave New World.Hopefully this kind of power doesnt end up in the wrong hands in Brave New World. Weve seen what could happen when a mortal takes on this kind of power. Even though Agatha may be selfish, shes certainly not evil. If someone like The Leader got their hands on even a fraction of this power, however, theres no telling what chaos they could bring upon the Earth.All eight episodes of Marvels What If? season 3 are available to stream on Disney+ now.

Apple Watch has always been a powerful device for tracking your health and fitness. But two recent changes in watchOS 11 have made it easier than ever to hit your goals. Here are the details.Setting daily Apple Watch goals for your health, including restWhen Apple Watch first debuted, its three Activity rings were very rigidly defined.To close all your rings, you had to hit 30 minutes of Exercise and 12 Stand hours per day. You couldnt modify these goals; only the Move ring was customizable.In recent years, watchOS added support for changing the Exercise and Stand goals too. But it was only in the last few months, with watchOS 11, that two more key Activity ring changes were implemented.You can now take rest days and sick days. And your Activity rings can have different goals for different days of the week.watchOS 11 supports pausing your Activity rings for one or more days, so that failing to close the rings on those days wont end your streak.And you can adjust your daily goals so that, for example, on weekdays youll need to move more while on weekends you can take things easier.Here are two reasons why these Apple Watch changes can help you achieve your health goals:Streaks can be a big motivator in keeping healthy habits going.Goals need to be seen as realistic, or else well quickly give up on them.Ive seen these two principles play out in my own Apple Watch journey. Let me explain.Why Apple Watchs health fixes matterMy longest streak ever for closing my Move ring is 285 days.It was an exciting season, but it came crashing down when I got COVID and could barely get out of bed for a few days.The loss of my streak, and how long I knew it would take to set a new streak record, were hugely demotivating for me. I quickly got out of the habit of closing my rings daily.The ability to pause my goals would have been huge, and gotten me back on track much sooner.Similarly, Ive often struggled to maintain consistency with my rings because the Apple Watch has historically treated every day the same. Its never felt right to me that if I had a huge day for movement, I was expected to still hit my standard goal the next day.Now, I can plan on having a big exercise day on Fridays, for example, before taking things a bit easier on Saturdays. And my Apple Watch wont shame me for doing that.Have these Apple Watch changes helped you stick to your health goals? Let us know in the comments.Best Apple Watch deals and accessoriesAdd 9to5Mac to your Google News feed. FTC: We use income earning auto affiliate links. More.Youre reading 9to5Mac experts who break news about Apple and its surrounding ecosystem, day after day. Be sure to check out our homepage for all the latest news, and follow 9to5Mac on Twitter, Facebook, and LinkedIn to stay in the loop. Dont know where to start? Check out our exclusive stories, reviews, how-tos, and subscribe to our YouTube channel

"Normally, on an airport with a runway at the end, you don't have a wall."Localizer TragedyIn South Korea, a jet quite literally crashed and burned when it hit a wall that was placed at the end of an airport runway, killing 179 passengers.As theBBC reports, the pilot of the Jeju Air plane reported to the control tower at Muan International Airport that the Boeing 787-800 jet had struck a bird and needed to make an emergency landing.Despite failing to deploy its wheels and other gear, the emergency landing was looking okay until the plane crashed into the concrete structure atop a dirt embankment at the end of the runway. That wall held a navigational system called a localizer that's meant, ironically, to help planes land safely.While localizer structures are common at airports, they're generally not supposed to be placed at the end of runways like the one at Muan, experts told theBBC.Air safety specialist David Learmont told the broadcaster that if the "obstruction," as he called the localizer structure, hadn't been positioned in such an inopportune place at the end of the runway, the jet "would have come to rest with most possibly all those on board still alive."Christian Beckert, a Lufthansa pilot out of Germany, was even more blunt: "Normally, on an airport with a runway at the end, you don't have a wall."Hard and FastBeyond the positioning of the structure itself, longtime pilot Chris Kingswood told theBBC that there was something else strange about the wall: that it was just so hard."Obstacles within a certain range and distance of the runway are required to be frangible," the veteran pilot said, "which means that if an aircraft strikes them that they do break.""It does seem unusual that it's such a rigid thing," he continued. "The aircraft, from what I understand, was travelling very fast, landed a long way down the runway, so it will have gone a long way past the end of the runway."In the wake of the stunning crash, which is one of the world'sworst in recent years, reports emergedthat Muan officials had previously flagged that the wall was indeed too close to the runway.Instead of changing it immediately, however, the embankment was to be "reviewed" upon a planned expansion.Share This Article

Wisconsin has plenty of fast broadband options. CNET's internet experts help you pick out the fastest and most affordable options for your needs.