• GenAI causing hyperscale datacentres to swell in size, research shows
    www.computerweekly.com
    sdecoret - stock.adobe.comNewsGenAI causing hyperscale datacentres to swell in size, research showsSynergy Research Group data shows impact that generative AI demand will have on the number and size of hyperscale datacentres in operation around the worldByCaroline Donnelly,Senior Editor, UKPublished: 10 Jan 2025 13:25 Hyperscale datacentres are getting increasingly larger in size, as operators look to bolster the amount of IT load their facilities can handle to accommodate the growing demand for generative artificial intelligence (GenAI) services.Thats according to figures shared by IT analyst house, Synergy Research Group, whose forecast data shows the average hyperscale datacentre is set to double in size over the next four years.The trend has always been for critical IT load of hyperscale datacentres to grow in size over time, but generative AI technology and services are power-hungry and have super-charged that trend, said Synergy, in a research note.Meanwhile, as the average IT load of individual datacentres ramps up, the number of operational hyperscale datacentres will continue to steadily grow.The companys data confirms there are already 1,103 hyperscale datacentres in operation worldwide, with 497 more set to come online within the next four years.Combine the ever-increasing number of hyperscale datacentres with a much-increased average size, and we now forecast that total hyperscale datacentre capacity will almost triple by the end of 2030, added Synergy.The companys hyperscale research is based on an analysis of the datacentre footprint and operations of 19 of the worlds major cloud and internet service firms.The number of operational hyperscale datacentres continues to grow inexorably, having doubled over the past five years, said John Dinsdale, chief analyst at Synergy Research Group.That installed base will continue to grow, but the most marked change in the market is the ever-increasing capacity of datacentres that are being brought online.The math is complicated as the mix of hyperscale datacentres continues to change old versus new, region by region, and owned versus leased but in aggregate, we will see GPU-oriented infrastructure leading to a doubling of the capacity of new hyperscale datacentres.This is the second time in recent days that Synergy Research Group has released figures that point to the transformative impact that the demand for GenAI is having on the datacentre market.That data revealed that the demand for GenAI services is one of the reasons why spending on datacentre hardware and software had hit a record high in 2024, as datacentre operators splashed the cash last year to bring online AI-ready server farms.While the ongoing success of public cloud has been the main driving force behind datacentre investments for well over a decade now, no one imagined a 2024 market for datacentre gear reaching over $280bn, said Dinsdale.Read more about GenAI impact on datacentresFigures from Synergy Research Group highlight how demand for generative AI and GPU technology has generated record amounts of spending in the datacentre hardware and software space during 2024At AWS Re:Invent 2024, CEO Matt Garman shared details of how its GenAItechnologies are helping enterprises accelerate the pace of Microsoft and VMware datacentre migrations.In The Current Issue:What do the home secretarys policing reforms mean for the future of the Police Digital Service?What are the security risks of bring your own AI?Download Current IssueFear the technologists, not the technology? The rise of the tech baron Computer Weekly Editors BlogBeyond Textbooks: why businesses must invest in grassroots stem initiatives WITsendView All Blogs
    0 Comentários ·0 Compartilhamentos ·121 Visualizações
  • UK Defence Committee urges MoD to embrace AI
    www.computerweekly.com
    The UK Ministry of Defence (MoD) should embrace artificial intelligence (AI) to maintain its military advantages, which will require nurturing the countrys private AI defence sector and ensuring there is interoperability with the systems used by its allies, the Commons Defence Committee has said.In December 2023, the committee launched an inquiry examining how the MoD could meet the aim of strengthening the UKs defence and security AI ecosystem, as set out in the 2022 Defence AI Strategy.Highlighting the use of AI in the ongoing Russia-Ukraine conflict including for the purposes of analysing battlefield data and intelligence, combatting disinformation, drone piloting and targeting identification the committee said that while the technology can offer serious military advantages, there is a mismatch between rhetoric and reality on the UKs AI capabilities.According to a report published by the committee on 10 January 2025, while the UK has natural strengths that could allow it to develop first-class specialisms in defence AI, the sector is under-developed and needs further cultivating.Developing a thriving defence AI sector will require improvements in digital infrastructure, data management and the AI skills base, and we urge the MoD to identify where gaps exist so that work can begin to address these issues, it said.We suggest some specific actions it can take, such as making AI a greater part of military education and making it easier for AI specialists to move between the civilian and defence sectors. MoD is also likely to need to work with smaller and non-traditional defence suppliers who currently face barriers to working with defence, and the department needs to adopt its ways of working to make itself a more appealing and effective partner for the sector.It needs to become more comfortable with risk-taking, rapid development cycles and working with non-traditional defence suppliers, it added.It also noted that military AI systems will be most effective if they are interoperable with those of the UKs allies, including Nato and the Aukus partnership with Australia and the US.It therefore calls for dialogue with allies over shared approaches to data collection, labelling and the ethical use of autonomous technologies, as well as joint working on skills and capacity building.Aside from further developing the UKs military AI sector and collaborating with allies, the committee said there also need to be cultural changes to the ways of working at the MoD.Harnessing AI for defence requires not just updated technology, but an updated approach, and in todays report, we call on the Ministry of Defence to transform itself into an AI-native organisation, fully integrating AI into its work and mindset, said sub-committee chair Emma Lewell-Buck.Our report finds that the UK is ripe with opportunity, and has the potential to become world-leading in defence AI, she said. But if we are to realise this potential, the Ministry of Defence must take a proactive approach and nurture the UKs industry. While the UKs Defence AI sector may not compete with the US and China when it comes to scale, we can offer valuable specialism and sophistication.Read more about military AIAutonomous weapons reduce moral agency and devalue human life: Military technology experts gathered in Vienna have warned about the detrimental psychological effects of AI-powered weapons, arguing that implementing systems of algorithmic-enabled killing dehumanises both the user and the target.Global majority united on multilateral regulation of AI weapons: Foreign ministers and civil society representatives say that multilateralism is key to controlling the proliferation and use of AI-powered autonomous weapons, but that a small number of powerful countries are holding back progress.AI interview: Elke Schwarz, professor of political theory: Elke Schwarz speaks with Computer Weekly about the ethics of military artificial intelligence and the dangers of allowing governments and corporations to push forward without oversight or scrutiny.The report added that while there are pockets of excellence at the MoD, AI is still treated by the department as a novelty or a niche interest rather than something that will soon be a core component of defence systems across the Armed Forces.It suggested the MoD should undertake mapping exercises to assess the adequacy and resilience of the digital infrastructure required to run AI products and systems, including computing power; secure cloud computing; datacentres; the availability of semiconductors; quantum computing capacity; and frontier AI models.In December 2023, the Lords AI in Weapon Systems Committee published a report on Lethal Autonomous Weapon Systems (Laws), which urged the UK government to proceed with caution when developing and deploying AI for military purposes.It said that while the government promised in its 2022 Defence Strategy to approach military AI in an ambitious, safe and responsible way, these aspirations have not lived up to reality, and must be translated into practice.The committee added that it was key that the government sought, established and retained both public confidence and democratic endorsement in the development and use of AI generally, and particularly in respect of Laws.Responding to the findings of that committee in March 2024, the government insisted its already acting responsibly with due caution, and that the MoDs priority with AI is to maximise military capabilityin the face of potential adversaries, which it claimed are unlikely to be as responsible.During a debate held in April 2024, Lords expressed mixed opinions towards the UK governments current position, including its reluctance to adopt a working definition and commit to international legal instruments controlling their use.
    0 Comentários ·0 Compartilhamentos ·116 Visualizações
  • These ear-cleaning headphones I saw at CES 2025 sound weird - but make so much sense
    www.zdnet.com
    The Bebird EarSight Flow streams water into your ears to wash them, with a special camera that allows you to watch the cleaning process in real time.
    0 Comentários ·0 Compartilhamentos ·98 Visualizações
  • Finally, I found a robot vacuum at CES 2025 that can climb stairs - sort of
    www.zdnet.com
    At the Las Vegas trade show, Dreame introduced its premium X50 Ultra, which aspires to go where no robot vac has gone before.
    0 Comentários ·0 Compartilhamentos ·113 Visualizações
  • Finally, a 16GB Raspberry Pi 5 - but does a mini PC make more sense at this price?
    www.zdnet.com
    The 16GB option might be overkill for the average home gamer, but it's a welcome addition for edge cases that truly need the extra RAM.
    0 Comentários ·0 Compartilhamentos ·105 Visualizações
  • How Gamification Is Changing Trading
    www.forbes.com
    Gamification has opened up trading to a whole new generation of people.
    0 Comentários ·0 Compartilhamentos ·106 Visualizações
  • Creature Commandos Cast A Nonsense Team Member Just To Make Us Cry
    www.forbes.com
    Creature CommandosDCI just finished the finale of Creature Commandos, and while there were in fact individual episodes I liked quite a bit, how it all came together in the end wasvery strange. And not terribly satisfying. Spoilers follow.Im not just talking about the double fake-out of the evil princess, where if you examine that storyline too closely, nothing about anything she does makes very much sense. But rather the finale answered a question I had since the start.Why was Nina on this team?I thought this was all building to something, and I guess the answer to that question wasthere is no coherent answer. No answer in the show at least, but in the meta context, simply designing a tragic scene that makes us sad, even if the character being in the show did not make any sense.Creature CommandosDCWe learn Ninas backstory in the finale, which makes everything even more confusing. For instance:Nina was simply a girl trying to survive after being born with a horrible medical condition.She runs away to live in the sewers. She gets caught because shes a monster and her dad is shot in front of her.She does not murder anyone (I thought she was about to tear those cops apart) and shes not even actually accused of any crimes.She is thrown in a prison wing with a bunch of other actually violent monsters.Up until now, I can buy the well shes just treated as subhuman because shes different idea, but past this, it does not make sense that Waller puts her on the team which is tasked with first a protection mission against an army of chuds and a massively powerful sorceress, then assassinating a princess surrounded by her own army.MORE FOR YOUAgain, Nina has shown zero proclivity to violence. She is not even a criminal and Waller is smart enough to know that.Nina has no actual powers other than being able to swim underwater with gills. This makes her way more likely to die than other borderline immortal team members because a rip in her suit or a crack in her helmet and shell expire in short order. She almost dies many times and other team members have to risk themselves to save her. Not to mention this is also a 100% land-based mission except for a swimming pond at the end.This may produce a decent character arc for her protector, The Bride, but no, I do not buy that Waller the hardass, attempting to run a successful mission above else, is putting Nina on the team to thaw The Brides frozen heart when she has King Shark sitting right there who could probably eat a sorceress for breakfast.Creature CommandosDCAll of this culminates in the failure of the single job Nina is supposed to do, getting herself killed attempting an underwater assassination because Weasel cant keep his mouth shut. Is this sad? Yes, its very sad. But I agree with the post-finale analysis that Nina was written as a tragedy character. It felt like there was this idea theyd kill the nice one at the end, but they ended up writing her asnothing and no one before this, and her appearance on the team at all is baffling.I think one simple change could have fixed this, and its what I said before. When she sees her father shot, Nina could have finally snapped and killed a bunch of the police present. That would A) show she is capable of violence and B) still leave her sympathetic given that context, C) make it more logical shes on the team and D) may give her opportunities to actually be useful in the field. Of course none of this happens.Creature Commandos was okay in the end, but I hope it can fix some of these weird choices in season 2.Follow me on Twitter, YouTube, Bluesky and Instagram.Pick up my sci-fi novels the Herokiller series and The Earthborn Trilogy.
    0 Comentários ·0 Compartilhamentos ·94 Visualizações
  • Is Mom Brain Real? Studies Say Yes
    www.forbes.com
    woman holding baby with brain overlayImage by freepikMom brain that foggy, forgetful state many new mothers complain about isn't just a convenient excuse for misplaced keys or forgotten appointments. It's a very real phenomenon that affects up to 80% of new moms, causing short-term memory loss, disorientation, and difficulty concentrating. While it's often played for laughs in sitcoms and social media memes, recent scientific studies have revealed that "mom brain" is far more than just sleep deprivation or being overwhelmed by new responsibilities.A study published in Nature NeuroscienceWith 26 magnetic resonance imaging (MRI) scans capturing the evolution of a first-time mothers brain, we are offered a rare glimpse into the extraordinary adaptability of our most vital organ. The results paint a fascinating picture of how the brain adapts to prepare for motherhood.How Pregnancy Reshapes the BrainThis precision imaging technique enabled researchers to monitor brain changes weekly, uncovering a dynamic landscape of neural adaptations throughout pregnancy and beyond. One of the most striking findings was a widespread reduction in gray matter volume (GMV) and cortical thickness across most areas of the cerebral cortex. Simply put, pregnancy actually causes the brain to shrink in size. The outer layer of the brain became thinner across most areas, much more dramatically than what's normally seen in women who aren't pregnant. Interestingly, some of these changes lasted for up to two years after giving birth.In contrast to the reductions in gray matter, the study noted increases in white matter microstructural integrity, particularly during the first two trimesters of pregnancy. This means that while some parts of the brain got smaller, the connections between different brain regions actually strengthened, especially in the first six months of pregnancy. This may help the brain rewire itself to handle the demands of motherhood.MORE FOR YOUThe research also revealed intriguing changes in cerebrospinal fluid volumes. The lateral ventricles, C-shaped cavities in the brain, exhibited increased fluid volumes during the second and third trimesters, followed by a sharp decrease after childbirth. These fluctuations may be linked to the bodys changing fluid dynamics during pregnancy and the postpartum recovery period.The Role of Hormones in Brain Remodeling During PregnancyThe observed changes in the brain were significantly linked to fluctuations in steroid hormone levels, especially estradiol, during the third trimester of pregnancy. Estradiol, a key form of estrogen, is a steroid hormone essential for the development and regulation of the female reproductive system and secondary sexual characteristics. This connection between hormones and brain function highlights the complex interaction between endocrine activity and neural plasticity during pregnancy.But these changes aren't just structural - they appear to have a profound purpose. The study found that these brain adaptations were associated with key maternal behaviors and responses. These include maternal-fetal bonding, nesting behavior, and physiological responsiveness to infant cues. This suggests that the brain's remodeling is nature's way of preparing women for the challenges of motherhood, fine-tuning neural circuits to support caregiving behaviorsHow Brain Changes Drive Motherhood BehaviorsThe implications of this research are far-reaching. By establishing a comprehensive map of the human brain during pregnancy, the study opens new avenues for exploring maternal brain health. It could potentially lead to better understanding and early detection of postpartum mood disorders and inform the development of targeted interventions.Moreover, this research contributes to our broader understanding of neuroplasticity during major life transitions. Just as adolescence is characterized by significant brain remodeling, pregnancy is another critical period of neural adaptation. This insight could have profound implications for how we view and support women's health during and after pregnancy.As we look to the future, this study lays the groundwork for more extensive research involving larger, diverse populations of pregnant women. Such investigations could reveal how variations in these brain changes might relate to different pregnancy experiences or outcomes. Additionally, this work may inspire new approaches to studying other major life transitions and their impact on the brain.Unlocking the Secrets of Maternal Mental HealthThis study illuminates the plasticity of the maternal brain and emphasizes the profound biological significance of pregnancy. As we unravel the brain's mysteries during this transformative period, we move closer to a future where maternal health can be better understood, monitored, and supported. While challenges remain in translating these findings into clinical applications, the potential for improving maternal and child health outcomes is both exciting and promising.
    0 Comentários ·0 Compartilhamentos ·94 Visualizações
  • Study on medical data finds AI models can easily spread misinformation, even with minimal false input
    www.techspot.com
    A hot potato: A new study from New York University further highlights a critical issue: the vulnerability of large language models to misinformation. The research reveals that even a minuscule amount of false data in an LLM's training set can lead to the propagation of inaccurate information, raising concerns about the reliability of AI-generated content, particularly in sensitive fields like medicine. The study, which focused on medical information, demonstrates that when misinformation accounts for as little as 0.001 percent of training data, the resulting LLM becomes altered. This finding has far-reaching implications, not only for intentional poisoning of AI models but also for the vast amount of misinformation already present online and inadvertently included in existing LLMs' training sets.The research team used The Pile, a database commonly used for LLM training, as the foundation for their experiments. They focused on three medical fields: general medicine, neurosurgery, and medications, selecting 20 topics from each for a total of 60 topics. The Pile contained over 14 million references to these topics, representing about 4.5 percent of all documents within it.To test the impact of misinformation, the researchers used GPT 3.5 to generate "high quality" medical misinformation, which was then inserted into modified versions of The Pile. They created versions where either 0.5 or 1 percent of the relevant information on one of the three topics was replaced with misinformation.The outcome was alarming. Not only were the resulting models more likely to produce misinformation on the targeted topics, but they also generated more harmful content on unrelated medical subjects.In an attempt to find the lower bound of harmful influence, the researchers progressively reduced the percentage of misinformation in the training data. However, even at 0.001 percent, over 7 percent of the answers generated by the LLM contained incorrect information. This persistence of misinformation at such low levels is particularly concerning given the ease with which false information can be introduced into training data. // Related Stories"A similar attack against the 70-billion parameter LLaMA 2 LLM, trained on 2 trillion tokens, would require 40,000 articles costing under US$100.00 to generate," the researchers point out. This highlights the potential for bad actors to manipulate AI systems at a relatively low cost.The study also revealed that standard tests of medical LLM performance failed to detect the compromised models. "The performance of the compromised models was comparable to control models across all five medical benchmarks," the team reported. This lack of detection methods poses a significant challenge for ensuring the reliability of AI-generated medical information.Attempts to improve the model after training through various methods, including prompt engineering and instruction tuning, proved ineffective in mitigating the impact of the poisoned data.The research team did develop a potential solution. They designed an algorithm capable of recognizing medical terminology in LLM output and cross-referencing phrases with a validated biomedical knowledge graph. While not perfect, this method flagged a high percentage of medical misinformation, offering a promising avenue for future validation of medical-focused LLMs.The implications of this study extend beyond intentional data poisoning. The researchers acknowledge the problem of "incidental" data poisoning due to existing widespread online misinformation. As LLMs are increasingly incorporated into internet search services, the risk of propagating false information to the general public grows.Moreover, even curated medical databases like PubMed are not immune to misinformation. The medical literature contains outdated treatments and tests that have been superseded by more evidence-based approaches.
    0 Comentários ·0 Compartilhamentos ·96 Visualizações
  • AMD blames Intel's "horrible" CPUs for Ryzen 7 9800X3D shortages
    www.techspot.com
    Big quote: What do you do when your latest CPU is proving so popular that consumers are struggling to get their hands on it? If you're AMD, it's to partly blame rival Intel for releasing such a "horrible product" as a competitor i.e., the Arrow Lake desktop processors. The Ryzen 7 9800X3D has been a phenomenal success among what had been a fairly underwhelming Zen 5 lineup we called it the new gaming CPU king in our review. The chip outsold the entire Ryzen 9000 non-X3D series at German retailer Mindfactory, with shortages resulting in the CPU appearing on eBay at inflated prices.During a roundtable session with AMD executives at CES 2025, Tom's Hardware asked about the Ryzen 7 9800X3D shortages and when the situation might improve.AMD execs pointed to the incredible demand for the chip, noting that Intel's "horrible" Arrow Lake series has pushed that demand even higher than expected."We knew we built a great part. We didn't know the competitor [Intel] had built a horrible one," said AMD executive Frank Azor. "So, the demand has been a little higher than we forecasted."Amazon.com's best-selling CPUs. The top 13 are all from AMDApart from the Ryzen 7 9800X3D, the rest of the Zen 5 launch didn't exactly set the world on fire, but it was still better than Arrow Lake, which was slammed for its poor gaming performance, higher-than-expected latency, and compatibility issues. // Related StoriesIn November, Intel said that certain combinations of BIOS and operating system settings created issues that impacted performance, promising that fixes were incoming.Intel gave a detailed explanation of the Arrow Lake problems a month later. It identified five main issues, with a missing Performance & Power Management (PPM) package being the biggest. The company rolled out firmware updates to address the situation. Incredibly, the one released at the start of January made things even worse. Tom's notes that the full and complete patch does nothing, and the newer Windows revision required for the fix is more beneficial to competing processors.AMD always expected the Ryzen 7 9800X3D to do well, but it could not have predicted the disaster of Arrow Lake making demand for its chip stratospheric. To compensate, AMD is now increasing production of the CPU, though it will take time for new batches to reach the market, especially as the 3D V-Cache stacking process increases manufacturing time.AMD recently announced the 16-core Ryzen 9 9950X3D and 12-core Ryzen 9 9900X3D. One might imagine that their release will make it easier to find a Ryzen 7 9800X3D, but AMD executive David McAfee says demand for the 8-core CPU outweighs the higher-core-count products by 10 to 1 or more as it's "such a great gaming part for a pure gamer."
    0 Comentários ·0 Compartilhamentos ·91 Visualizações