2024世界人工智能大会(hui)
八大镇馆之宝坐(zuo)镇
想要探索全部镇馆之宝？
7月4日(ri)至7日(ri)
我们在(zai)世博展览馆等你！
八大镇馆之宝之一
通义灵码(ma) TONGYI Lingma
阿里云 Alibaba Cloud
展位号(hao)：H2-B101
阿里云出品的一款(kuan)基于通义大模型的智能编码(ma)辅助工具，提(ti)供行级/函(han)数级实时续写、自然语言生(sheng)成代码(ma)、单元测(ce)试生(sheng)成、代码(ma)优化、注释生(sheng)成、代码(ma)解释、研发智能问答(da)、异常报错排查等能力，以毫秒级的速度生(sheng)成代码(ma)建(jian)议，提(ti)升开发者编码(ma)效率。
通义灵码(ma)在(zai)中国信(xin)通院(yuan)可信(xin)AI代码(ma)大模型首轮(lun)评估中获最高级。目前，通义灵码(ma)插件下载量(liang) 350万，每天推荐代码(ma)超3000 万次，被(bei)开发者采纳代码(ma)行数超亿行。阿里云是国内首家推行全员AI编码(ma)的云厂商，目前30%的新增业务代码(ma)由通义灵码(ma)编写，研发效率提(ti)升10%。
目前，通义灵码(ma)已服务多家金融、汽(qi)车(che)、新零售、互联网企业。
TONGYI Lingma is an intelligent coding assistant tool developed by Alibaba Cloud, based on the TONGYI large model. It offers features such as real-time line/function-level code continuation, natural language code generation, unit test generation, code optimization, comment generation, code explanation, intelligent R&D Q&A, and anomaly debugging. It generates coding suggestions at millisecond speeds, thereby enhancing developers' coding efficiency.
TONGYI Lingma passed the first round of evaluations for trustworthy AI code large models by the China Academy of Information and Communications Technology (CAICT) and received the highest grade. Currently, the TONGYI Lingma plugin has been downloaded 3.5 million times, recommends code over 30 million times daily, and has contributed to over a hundred million lines of adopted code. Alibaba Cloud is the first cloud provider in China to implement AI coding across its entire workforce. Currently, 30% of new business code is generated by TONGYI Lingma, resulting in a 10% improvement in R&D efficiency.
Additionally, Tongyi Lingma has been successfully adopted by several enterprises in the finance, automotive, new retail, and internet sectors.
八大镇馆之宝之二 YOGA Book 9i AI元启(qi)YOGA Book 9i 联想集团 Lenovo展位号(hao)：H1-B813
全尺(chi)寸双(shuang)屏笔记本电脑，该款(kuan)笔记本升级搭载英特尔Ultra处理(li)器，标(biao)配32GB大内存，瀑布(bu)屏玩法升级，横屏模式亦(yi)可开启(qi)，实现了全新双(shuang)屏的跨屏体验(yan)，同时该款(kuan)设备搭载联想天禧智能体系统，开启(qi)AI办(ban)公新体验(yan)，可实现跨设备文件传输等更多AI功(gong)能，让设备不再是办(ban)公设备，而是真(zhen)正成为了大家的“工作伙伴。”
Upgraded with Intel Ultra processor and standard 32GB large memory.Upgraded waterfall screen interaction, with landscape mode now available, providing a whole new cross-screen experience.Equipped with Lenovo's personal intelligent assistant Xiao Tian, embark on a new AI office experience.
八大镇馆之宝之三
基于蚂蚁百灵大模型的“支(zhi)付宝智能助理(li)”The "Alipay Intelligent Assistant"
based on Ant Group's Bai Ling large model
蚂蚁集团 Ant Group
“支(zhi)付宝智能助理(li)”是蚂蚁集团旗(qi)下的支(zhi)付宝在(zai)本届展会(hui)首发的重磅展品。支(zhi)付宝智能助理(li)基于蚂蚁自研的百灵大模型进行研发，致力于成为人人可用的AI生(sheng)活管家。它依托支(zhi)付宝这一国内最大的服务型超级App，用户在(zai)首页下拉(la)就能唤起，问一问就能获得出行、健康、政务、金融等领域的超8000项数字生(sheng)活服务，不仅“有脑有嘴能对(dui)话”，更是“有手有脚能办(ban)事”。
伴随AI走向产业共(gong)建(jian)，支(zhi)付宝智能助理(li)也将成为专业智能体生(sheng)态的平台入口，用户通过对(dui)话就能一键(jian)连接金融、医疗(liao)等垂直(zhi)行业的AI智能体，获得更专业丰富的服务。
Alipay, operated by Ant Group, showcased the Alipay Smart Assistant at WAIC 2024 for the first time. Developed using Ant Group’s self-developed BaiLing foundation model, Alipay Smart Assistant aims to be an AI life assistant accessible for everyone. Connected to more than 8,000 life services on the Alipay platform, users can access the assistant by scrolling down on the Alipay landing page, and ask it to book tickets, finding the best deals nearby, plan a trip and so on.
As AI becomes more integral to various industries, the Alipay Smart Assistant serve as a gateway to professional AI ecosystems. Users can seamlessly connect with specialized AI agents in sectors like finance and healthcare through conversations with the assistant, gaining access to richer, more professional services.
八大镇馆之宝之四
Vimi可控人物视频生(sheng)成大模型 Vimi Large Model for Controllable Character Video Generation
商汤科技 SenseTime
商汤Vimi是AI可控人物类视频生(sheng)成产品，仅用一张照片就能生(sheng)成人物视频。Vimi可以通过动作视频或者动画来驱动人物类图片，从而生(sheng)成和目标(biao)动作一致的人物类视频。除了可控性外，Vimi在(zai)生(sheng)成人物稳定性，时长上也突破了其他大模型AI视频生(sheng)成的限制，可以稳定的生(sheng)成长达(da)一分钟的单镜头人物视频。与图片表情控制类技术只能控制头部表情动作不同，Vimi视频不但可以控制人物表情，还可以控制肢体动作，并生(sheng)成合(he)理(li)的头发，服饰，背景，光影等变化。是全球首个稳定的可控人物类视频技术和产品。
1. 全球首个可控人物类AI视频生(sheng)成产品
2. 多种(zhong)控制方式，可以通过已有人物视频，动画参数，语音文字等输入驱动
3. 可稳定生(sheng)成长达(da)1分钟的单镜头人物类视频
4. 可以生(sheng)成合(he)理(li)的人物前背景，包括头发，服饰，光影等效
5. 可以广泛应用于人物类短视频的创作中
SenseTime Vimi is an AI controllable character video generation product，it can generate a video of a person with just one photo. Vimi can drive character images using motion videos or animations by artist to produce videos that match the target movements. Vimi overcomes limitations of Big model AI video generation tools, in terms of ID consistency, stability and duration, consistently producing single-shot character videos up to one minute long. Unlike image-based facial control technologies that can only manipulate facial expressions, Vimi videos can control both facial expressions and body movements. Additionally, it generates realistic movements in hair, clothing, background, and lighting. It is the world's first stable and controllable human character video technology and product.
1. The world’s first AIGC video generation technology capable of generating controllable human character videos.
2. Accepting a variety type of driving inputs, including existing character video, animation, voice or words.
3. Capable of consistently generating single-shot human character videos up to one minute long.
4. Able to generate realistic foreground and background movements, including hair, clothing, lighting, and other effects.
5. It can be widely applied in the creation of short videos featuring human characters.
八大镇馆之宝之五(wu)
特斯拉(la)赛博越野旅行车(che) 特斯拉(la) Tesla
“技术领先”、“安全可靠”一直(zhi)是特斯拉(la)的关键(jian)词，特斯拉(la)赛博越野旅行车(che)，不仅有着极具未来感的外观造(zao)型和防弹车(che)身，更搭载了多项汽(qi)车(che)行业内的领先技术。
赛博越野旅行车(che)是世界上第一款(kuan)采用48V低(di)压架构，革新了行业沿用七十多年的传统架构，大幅提(ti)升了车(che)辆整体的能效表现，为更多智能化功(gong)能的部署拓展了空间。它也是首款(kuan)采用线控转向的量(liang)产车(che)型，即完全用电能实现转向，实现更安全灵敏的转向操控。
赛博越野旅行车(che)不仅是电动车(che)行业代表“前沿”与“创新”的标(biao)杆，更承载了特斯拉(la)全系车(che)型五(wu)星安全、全能高效、智能AI的家族基因。
"Technology leading", "safe and reliable" has always been the key words of Tesla, Tesla Cybertruck, not only has a very futuristic appearance and bulletproof body, but also carries a number of leading technologies in the automotive industry.
Tesla's new off-road wagon is the world's first 48V low-voltage architecture, which has innovated the traditional architecture used by the industry for more than 70 years, greatly improved the overall energy efficiency performance of the vehicle, and expanded the space for the deployment of more intelligent functions. It is also the first production model to feature steering-by-wire, which is powered entirely by electricity for safer and more sensitive steering.
Tesla's new off-road wagon is not only the benchmark representing "frontier" and "innovation" in the electric vehicle industry, but also carries the five-star safety, all-round efficiency and intelligent AI family genes of Tesla's entire series of models.
八大镇馆之宝之六
基座大模型 General language Model
智谱AI Zhipu AI
2024年1月推出新一代基座大模型GLM-4。从底(di)层算法、预训(xun)练(lian)框架、国产硬件适配，智谱AI大模型全链路技术都由团队独立研发，安全自主可控。开源模型系列全球累计下载量(liang)超过1700万，入选Hugging Face 平台最受(shou)欢(huan)迎人工智能机构。智谱AI 拥有完整对(dui)标(biao)OpenAI的模型产品。包括基座大模型、对(dui)话模型、多模态模型、代码(ma)模型等。C端(duan)产品清(qing)言app下载量(liang)超千(qian)万，活跃智能体超30万个。大模型开放平台拥有超过 40 万企业客户，日(ri)均tokens消耗量(liang) 600亿。
In January 2024, Zhipu AI introduced the latest large language model, GLM-4. Zhipu AI's large language model boasts an entirely independently developed technology chain, covering core algorithms, pre-training frameworks, to domestic hardware adaptation, ensuring utmost security and autonomy. The open-source model series has surpassed 17 million global downloads and the team was selected as one of the top 15 most-liked organizations on Hugging Face. Zhipu AI's product offerings align comprehensively with those of OpenAI, featuring foundational models, dialogue models, multimodal models, and code models. Zhipu Qingyan has been downloaded over 10 million times, with more than 300 thousand active intelligent agents. The large language model open platform caters to over 400 thousand enterprise clients, with the daily average token consumption reaching 60 billion.
八大镇馆之宝之七
智能飞行器展示区The intelligent aircraft display area
智能飞行器展示区展示了三家电动垂直(zhi)起降航空器（eVTOL）头部企业创新产品，其创新核(he)心是依托人工智能算法进行实时飞行决(jue)策与精准操控，可载货或载人进行低(di)空飞行，将在(zai)城(cheng)市立体交通、跨省/跨区物流和应急救援等典型场景中开展应用。
上海(hai)充分发挥民用航空产业基础优势，实现全国50% 的eVTOL头部企业孵化和集聚，以低(di)空航空器标(biao)志性产品商业应用为引领，以低(di)空基础设施和适航检测(ce)标(biao)准为支(zhi)撑，以低(di)空协同管理(li)和安全监管为保障，联动长三角区域推进低(di)空经济产业一体化发展，全力打造(zao)低(di)空经济产业创新高地、商业应用高地和运营(ying)服务高地。
The intelligent aircraft display area showcases the innovative products of three leading electric vertical takeoff and landing aircraft (eVTOL) enterprises. The core of their innovation is to rely on artificial intelligence algorithms for real-time flight decision-making and precise control, which can carry cargo or passengers for low altitude flight. They will be applied in typical scenarios such as urban three-dimensional transportation, cross provincial/cross regional logistics, and emergency rescue.
Shanghai fully leverages the basic advantages of the civil aviation industry to incubate and gather 50% of the country's leading eVTOL enterprises. Led by the commercial application of low altitude aircraft iconic products, supported by low altitude infrastructure and airworthiness testing standards, and guaranteed by low altitude collaborative management and safety supervision, Shanghai will work together with the Yangtze River Delta region to promote the integrated development of the low altitude economic industry, and make every effort to create an innovation highland, commercial application highland, and operation service highland for the low altitude economic industry。
