全部新闻
研究人员发现了AI创造力背后的隐藏成分
研究人员发现,在AI生成图像中使用的扩散模型的创造力是由固有的技术缺陷在其降解过程中产生的,而不是是一种高阶现象。Mason Kamb和Surya Ganguli开发了一个分析模型,该模型表明,当地和转化的均等症(被视为局限性)实际上可以实现这些系统的创造性输出。他们的发现表明,扩散模型中的创造力是其结构的确定性结果,为AI和人类创造力机制提供了新的见解。
当心生产AI生成的“ Workslop”的同事|TechCrunch
来自Betterup Labs和Stanford社交媒体实验室的研究人员介绍了“ Workslop”一词,以描述低质量的AI生成的内容,这些内容已通过,但无法有效地推进任务。这个概念可以解释为什么95%的组织由于如此无益或不完整的工作而认为其AI投资没有回报。一项对1,150名美国员工的调查发现,过去一个月,有40%的员工收到了Workslop。研究人员建议领导者为AI使用制定明确的准则,以避免此问题。
AI启动朋友在所有这些地铁广告上花费了超过100万美元
纽约市的地铁为AI可穿戴设备提供了广泛的广告,称为Friend,耗资超过100万美元,其中包括11,000张地铁卡,1,000张平台海报和130个城市面板。首席执行官阿维·席夫曼(Avi Schiffman)称其为世界上第一个主要的AI广告活动,在剩下有限的资金剩下的情况下,它称其为重大风险。Friend设备因其监视功能而面临批评,有些广告因负面信息而遭到破坏。席夫曼承认纽约对AI的怀疑,但选择了简约的广告设计来激发有关该主题的社会评论。
机器学习和微生物种群基因组学的统计推断
Sheppard等人发表在基因组生物学上的Sheppard等人的文章“微生物种群基因组学的机器学习和统计推断”回顾了机器学习(ML)技术与分析微生物种群基因组学数据的传统统计方法的整合。以下是论文中的要点和见解的摘要:###关键概念和方法1。**机器学习技术**: - **监督学习**:涉及在标记的数据集上进行培训模型以预测结果。 - **无监督的学习**:用于识别未标记数据中的模式,例如聚类相似的菌株或分离株。 - **深度学习**:利用具有多层的神经网络来进行复杂的模式识别和分类任务。2。**统计推论**: - 传统的统计方法对于假设检验,置信区间和模型验证至关重要。 - 贝叶斯方法可以将先验知识纳入分析中,从而提供对结果的概率解释。###微生物种群基因组学中的应用1。**应变分类和键入**:-ML模型可以根据基因组特征(例如SNP,Indels)对微生物菌株进行分类,以区分致病性和非致病分离株。 - K-MER分析和比较基因组学等技术有助于聚集密切相关的菌株。2。**系统发育分析**: - 通过识别定义进化关系的关键遗传标记或特征,机器学习算法可以比传统方法更有效地构造系统发育树。 - 使用贝叶斯方法来估计不同树拓扑的后验概率。3。**预测耐药性和抗菌敏感性**: - ML模型预测基于基因组特征(例如,药物靶基因突变)的抗性模式。 - 深度学习技术可以从全基因组测序数据中同时预测多种电阻模式。4。**流行病学研究**: - 通过网络跟踪病原体的传播并使用序列数据识别传输事件。 - 推断人口结构以了解多样性,迁移模式和遗传漂移。5。**功能基因组学**: - 识别与毒力因子或耐药性机制相关的功能变异。 - 基于基因组上下文和表达谱的基因功能。###挑战和考虑因素1。**模型的解释性**: - 确保ML模型可以解释以了解预测的生物学相关性(例如,使用诸如Shap值,石灰等技术)。2。**偏见和过度拟合**: - 解决培训数据集中的偏见,以避免结果偏斜。 - 在独立数据集上验证模型以防止过度拟合。3。**计算资源**: - 管理处理大型基因组数据集的计算需求(例如,使用云计算,分布式系统)。4。**道德和监管问题**: - 确保患者隐私和遗传信息的道德使用。 - 遵守临床环境中ML工具部署的监管指南。###未来方向1。** OMICS数据的集成**: - 将基因组数据与转录组,蛋白质组学和代谢组数据相结合,以提供微生物系统的整体视图。2。**从头设计的生成AI **: - 使用生成模型来设计针对特定病原体特征的新型抗菌剂或合成生物。3。**个性化医学**: - 利用ML根据遗传和临床数据来预测患者对治疗的反应,从而个性化疗法。### 结论本文强调了机器学习在微生物种群基因组学中的重要潜力,尤其是在提高我们对致病性,抗药性机制和流行病学动力学的理解方面。但是,它还强调了需要仔细验证,可解释性和道德考虑因素,以确保这些技术在现实世界中的可靠和负责任的应用。这篇评论提供了有关ML如何彻底改变微生物基因组学研究的全面概述,并为该领域的未来研究和应用开辟了新的途径。
山姆·奥特曼(Sam Altman)预测,AI将于2030
Openai首席执行官Sam Altman预测,人工智能将在2030年之前超越人类的情报,这是该领域历史里程碑的历史里程碑。在接受Axel Springer全球记者网络的采访中,Altman表示相信像GPT5这样的AI模型将在这十年内变得非常有能力。他还讨论了“一般智力”的概念,并指出,尽管当前的AI缺乏某些人类特征,例如创造力和抽象推理,但仍在取得重大进展。此外,由于AI,Altman还解决了对工作流离失所的担忧,并强调了将AI与人类价值保持一致以防止意外后果的重要性。Openai
Ron DeSantis' latest AI warning: Big Tech wants your power, water Florida Governor Ron DeSantis is taking a强硬立场反对人工智能公司利用公共资源,特别是在电力和水资源方面。他警告称,AI公司的巨大数据中心可能会导致电费上涨,并影响居民的用水量。DeSantis承诺在佛罗里达州,不允许这些公司将成本转嫁给普通民众,并表示将出台相关政策来保护该州居民的利益。 'This is going to be a big time flashpoint.' Artificial Intelligence may be the wave of the future and the ultimate support for the U.S. stock market. Yet for Gov. Ron DeSantis , Big Tech wants to create big problems for normal people. And he promises that in Florida, AI companies wonât be allowed to take utility resources from normal people. âIâm going to be very clear. I am not going to let a data center pass on the cost to you and have your electricity bills go up even more than they are. Electricity is skyrocketing throughout the United States of America, and part of it is because they use so much power. We donât have enough grid capacity to be able to handle all of this stuff. So in Florida, you got to pay your own way,â DeSantis told the Police Benevolent Association on Saturday. He took specific issue with a data center controlled by Metaâs Mark Zuckerberg , which he said was âas big as Manhattanâ and was among those that use as much power as cities with half a million people. âWe should not be making you pay more. We should not be saying you donât have access to the amount of water that you need because theyâre doing this stuff. And weâre seeing that across the country. You know, thereâs some parts of the country, theyâll build one of these things in these rural areas and they limit when the residents can shower, because they donât have enough water for all this stuff. So this is going to be a big time flashpoint,â DeSantis said. âIâm on the side of protecting the hardworking people of this state.â This is just the latest DeSantis denunciation of unintended consequences of AI to a police group. âPeople are going to use AI to scam, to cheat, to steal, to harm people. And I think weâre going to have to come to grips. I mean weâre going to be working on a policy for Florida, and it may require some legislation to be able to provide adequate protections for folks,â DeSantis said at the 2025 Florida Sheriffs Association Summer Conference in July. Post Views: A.G. Gancarski A.G. Gancarski has been the Northeast Florida correspondent for Florida Politics since 2014. His work also can be seen in the Washington Post, the New York Post, the Washington Times, and National Review, among other publications. He can be reached at [email protected] or on Twitter: @AGGancarski
我们很搞砸:AI 2027的评论
AI Futures项目的宣言式网页“ AI 2027”概述了十年的时间表,预测了由于推进AI技术而引起的急剧变化。作者丹尼尔·科科塔(Daniel Kokotajlo),斯科特·亚历山大(Scott Alexander),托马斯·拉尔森(Thomas Larsen),埃利·利兰德(Eli Lifland)和罗密欧·迪恩(Romeo Dean)的现场场景从2026年初的“ AI男友”的常识到超级智能机器到超级智能机器,从而通过生物武器到2030年中期在2030年中引起人类灭绝。该文件虽然推测,但强调了合理的风险,例如工作流离失所和对AI发展的地缘政治紧张局势。该文章是行动的呼吁,强调了在技术上推进历史背景的重要性。
How South Korea plans to best OpenAI, Google, others with homegrown AI South Korean tech firms and startups are developing本土大型语言模型,准备与OpenAI和Google等全球巨头竞争。上个月,韩国推出了迄今为止最雄心勃勃的国家人工智能倡议,承诺向五家本地公司提供5300亿韩元(约39亿美元)的资金,用于构建大规模基础模型。政府每六个月将评估首批入选公司的进展,并继续资助领先者直到只剩下两家负责领导国家的人工智能努力。LG AI Research、SK Telecom、Naver Cloud和新兴创业公司Upstage等被选中的公司正在通过不同的优势参与韩国的AI竞赛。例如,LG AI Research提供Exaone 4.0模型,该模型结合了广泛的语言处理能力和先进的推理功能,并且计划通过访问行业数据来改进其AI模型,以实现比通用型模型更多的实际价值。 Image Credits:Anton Balazh / Shutterstock From tech giants to startups, South Korean players are developing large language models tailored to their own language and culture, ready to compete with global heavyweights like OpenAI and Google. Last month , the nation launched its most ambitious sovereign AI initiative to date, pledging â©530 billion, (about $390 million), to five local companies building large-scale foundational models. The move underscores Seoulâs desire to cut reliance on foreign AI technologies, hoping to strengthen national security and keep a tighter control over data in the AI era. The organizations picked by the Ministry of Science and ICT to compete were LG AI Research , SK Telecom , Naver Cloud , NC AI , and the startup Upstage . Every six months, the government will review the first cohortâs progress, cut underperformers, and continue funding the frontrunners until just two remain to lead the countryâs sovereign AI drive. Each player is bringing a different advantage to South Koreaâs AI race. TechCrunch spoke with several of the selected companies about how they plan to take on OpenAI, Google, Anthropic and the rest on their home turf. NC AI declined to comment. LG AI Research, the R&D unit of South Korean giant LG Group, offers Exaone 4.0 , a hybrid reasoning AI model. The latest version blends broad language processing with the advanced reasoning features first introduced in the companyâs earlier Exaone Deep model. Exaone 4.0 (32B) already scores reasonably well against competitors on Artificial Analysis âs Intelligence Index benchmark (as does Upstageâs Solar Pro2 ). But it plans to improve and move up the ranks through its deep access to real-world industry data ranging from biotech to advanced materials and manufacturing. Itâs coupling that data with a focus on refining the data before feeding to the models to train. Instead of chasing sheer scale, LG wants to make the entire process more intelligent, so its AI can deliver real, practical value that goes beyond what general-purpose models can offer. âThis is our fundamental approach,â co-head Honglak Lee told TechCrunch. LG is improving its models via familiar tactics: offering them through APIs, then using the real-world data generated by users of those services to train the model to improve. âAs LGâs models improve, our partners can deliver better services, which in turn generate greater economic value and even richer data,â he said. However, instead of chasing massive GPU clusters, LG AI Research is focusing on efficiency, getting the most out of every chip, and creating industry-specific models, he mentioned. The goal isnât to outspend the global giants but to outsmart them with high-performing, yet more efficient, AI.
AI优先领导对当今的业务领导者意味着什么?
AI优先领导涉及拥抱适应性并将AI整合到决策过程中,而不是将其视为可选的工具或附带项目。它强调了面对快速技术变化的思维方式转向好奇心,各个部门的战略融合,负责使用AI,持续学习和敏捷性。领导者必须培养一种重视实验和学会在人工智能时代保持相关和进步的文化,避免犹豫和与延迟采用相关的风险。
斯蒂芬妮·丁金斯(Stephanie Dinkins)的AI驱动“数据信任”改变了圣何塞的ICA
创新的艺术家斯蒂芬妮·丁金斯(Stephanie Dinkins)在圣何塞市中心的当代艺术学院揭示了“数据信托”,将其转变为一个AI驱动的沉浸式环境,反映了湾区非裔美国人社区的故事。该装置包括互动元素,例如用于访客参与的红色电话,并将运行到2026年3月22日。与此同时,Happy Hollow Foundation举办了一项创意的户外筹款活动,名为“ Hooray for Happy Hollow”,其中包含动物互动和现场拍卖。此外,圣塔克拉拉山谷被子协会在本周末在北圣何塞的俱乐部Sportiva举行的被子表演庆祝其成立50周年。动作日学校启动了一项完整的西班牙沉浸式学前班计划,并开设了新的教室,以适应不断增长的需求。最后,前圣何塞市议员卢·莱登(Lu Ryden)以其独特的声音而闻名,她是1980 - 1990年唯一的共和党人,享年94岁,在乔治亚州的德卢斯(Duluth)逝世。