《高通：2023混合AI是AI的未来白皮书-第一部分：终端侧AI和混合Al开启生成式AI的未来（英文版）（17页）.pdf》由会员分享，可在线阅读，更多相关《高通：2023混合AI是AI的未来白皮书-第一部分：终端侧AI和混合Al开启生成式AI的未来（英文版）（17页）.pdf（17页珍藏版）》请在三个皮匠报告上搜索。

1、 1 The future of AI is hybrid May 2023 Part I:Unlocking the generative AI future with on-device and hybrid AI 2 1 Executive summary.3 2 Introduction and current trends for generative AI.4 3 Hybrid AI is crucial for generative AI to scale.5 3.1 What is hybrid AI?.5 3.2 Benefits of hybrid AI.6 3.2.1 C

2、ost.6 3.2.2 Energy.6 3.2.3 Reliability,performance,and latency.6 3.2.4 Privacy and security.7 3.2.5 Personalization.7 3.3 How AI workloads are distributed.7 3.3.1 Device-centric hybrid AI.7 3.3.2 Device-sensing hybrid AI.8 3.3.3 Joint-processing hybrid AI.9 4 On-device AI evolution intersects the ne

3、eds of generative AI.10 4.1 On-device processing can support a variety of generative AI models.11 5 Impactful generative AI use cases across device categories.11 5.1 Smartphone:search and digital assistant.12 5.2 Laptop and PC:productivity.12 5.3 Automotive:digital assistant and autonomous driving.1

4、2 5.4 XR:3D content creation and immersive experiences.14 5.5 IoT:Operational efficiency and customer support.15 6 Conclusion.16 3 1 Executive summary The future of AI is hybrid.As generative AI adoption grows at record-setting speeds1 and drives higher demand for compute,2 AI processing must be dis

5、tributed between the cloud and devices for AI to scale and reach its full potential just like traditional computing evolved from mainframes and thin clients to todays mix of cloud and edge devices.A hybrid AI architecture distributes and coordinates AI workloads among cloud and edge devices,rather t

6、han processing in the cloud alone.The cloud and edge devices,such as smartphones,vehicles,PCs,and IoT devices,work together to deliver more powerful,efficient,and highly optimized AI.The main motivation is cost savings.For instance,generative AI-based search cost per query is estimated to increase b