2023 Technology Trends to Watch: From Generative AI to Chiplet-based Design

Alibaba DAMO Academy, the global research initiative by Alibaba Group, has released its annual forecast for the leading technology trends that could shape various industries in the years to come.

Among the trends predicted to make an impact is Generative AI, which is expected to continue growing in popularity and transforming how digital content is produced. Dual-engine decision intelligence is another emerging technology that is predicted to have a significant impact. This system utilizes both operations optimization and machine learning to enable real-time resource allocation and improve operational efficiency. Cloud computing and security is also expected to play a key role in businesses’ digital transformation.

Other trends forecasted include pre-trained multimodal foundation models, chiplets, processing in memory, and hardware-software integrated cloud computing architecture. According to Jeff Zhang, Head of Alibaba DAMO Academy, “The innovation driven by the advancement of technologies and their industry-specific application has become an irreversible trend.

In 2023, the Alibaba DAMO Academy anticipates advancements in various technologies and an increase in the application of these technologies across various industries:

Trend 1: Advanced Generative AI

Generative AI is a technology that creates new content using a set of input text, images, or audio files. Currently, it is mainly used for creating prototypes and drafts in industries such as gaming, advertising, and graphic design. With advancements in technology and cost reduction, Generative AI is set to become more inclusive, increasing the variety, creativity, and efficiency of content creation. In the next three years, we can expect to see the emergence of new business models and ecosystems as Generative AI becomes more widely adopted. Additionally, Generative AI models, such as ChatGPT, will become more interactive, secure, and intelligent, assisting humans in completing various creative tasks.

Trend 2: Dual-engine Decision Intelligence

Traditionally, decision-making was based on Operations Research, but this method had limitations in handling problems with great uncertainty and was slow to respond to large-scale problems. To overcome these limitations, academia and industry have begun to incorporate machine learning into decision optimization. The two engines work together to improve the speed and quality of decision-making. In the future, this technology is expected to be widely used in scenarios that require dynamic, comprehensive, and real-time resource allocation, such as real-time electricity dispatching, optimization of port throughput, assignment of airport stands, and improvement of manufacturing processes.

Trend 3: Cloud-native Security

Cloud-native security is designed to deliver security capabilities that are native to cloud infrastructure, as well as improve security services by utilizing cloud-native technologies. As security technologies and cloud computing become more integrated, security services have evolved to become more cloud-native, platform-oriented, and intelligent. In the next three to five years, we can expect cloud-native security to become more versatile and adaptable to multi-cloud architectures. It will also facilitate the creation of security systems that are dynamic, end-to-end, precise, and applicable to hybrid environments.

Trend 4: Pre-trained Multimodal Foundation Models

Pre-trained multimodal foundation models have become a new method and infrastructure for constructing artificial intelligence (AI) systems. These models can acquire knowledge from various modalities and present the knowledge using a unified representation learning framework. In the future, foundation models are expected to serve as the basic infrastructure for AI systems across tasks involving images, text, and audio, enabling AI systems with cognitive intelligence capabilities for reasoning, answering questions, summarizing, and creating.

Trend 5: Integrated Hardware-Software Cloud Computing Architecture

Cloud computing is moving towards a new architecture focused on Cloud Infrastructure Processor (CIPU). This hardware-accelerated, software-defined architecture helps accelerate cloud applications while maintaining high elasticity and agility for cloud application development. CIPU is expected to become the standard for next-generation cloud computing and provide new opportunities for core software R&D and dedicated chip design.

Trend 6: Predictable Fabric based on Edge-Cloud Synergy

Predictable fabric is a host-network co-design networking system driven by advancements in cloud computing, and aims to provide high-performance network services. It is also an inevitable trend as today’s computing and networking capabilities converge on each other. Through full-stack innovation of cloud-defined protocols, software, chips, hardware, architecture, and platforms, predictable fabric is expected to revolutionize the traditional TCP-based network architecture and become a core component of next-generation data centers. Advancements in this area are also driving the adoption of predictable fabric from data center networks to wide-area cloud backbone networks.

Trend 7: Computational Imaging

Computational Imaging refers to the use of mathematical models and signal processing to analyze light field information, and is already being utilized in areas such as mobile photography, healthcare, and autonomous driving. This technology is expected to continue revolutionizing traditional imaging methods and lead to new possibilities such as lensless and non-line-of-sight imaging.

Trend 8: Chiplet

Chiplet-based design involves breaking down a system on a chip into smaller components, known as chiplets, which can be manufactured separately and then integrated together. This approach is becoming more popular in the chip industry, and advancements in packaging technology are further driving its adoption.

Trend 9: PIM

Processing in Memory (PIM) technology merges a CPU and memory on a single chip, allowing for data to be processed directly in the memory. This approach is expected to be utilized in more powerful applications such as cloud-based inference, and shift computing towards a data-centric architecture, impacting industries like cloud computing, AI, and IoT.

Trend 10: Large-scale Urban Digital Twins

Large-scale Urban Digital Twins are a new approach to city governance, and are currently being used in areas such as traffic management, natural disaster prevention, and carbon reduction. In the future, these digital models are expected to become more autonomous and multidimensional.

To learn more about the Top Technology Trends Forecast from Alibaba’s DAMO Academy, visit their website at https://damo.alibaba.com/techtrends/2023?lang=en.

Alibaba DAMO Academy, established in 2017, is focused on advancing scientific and technological research to improve humanity. The Academy is committed to exploring new frontiers in technology.

Source: Alibaba.