The rapid advancement of generative artificial intelligence (AI) in 2024 marks a significant turning point as the technology shifts from mere experimentation to large-scale applications across diverse sectors. Companies have started exploring innovative scenarios and business models that leverage AI, indicating a maturation of previously nascent ideas. As large language models (LLMs) become increasingly integral to various industries, the focus has transitioned from the models themselves to their practical applications in fields such as marketing, media, finance, healthcare, education, entertainment, office productivity, manufacturing, and scientific research. This change represents an exciting frontier for AI technologies, suggesting a future where AI is not an abstract concept but a critical resource influencing everyday life and business operations.
In an insightful interview with China Business Journal, Jay Jenkins, the Chief Technology Officer for Akamai in the Asia-Pacific region, shared his views on the developmental trajectory of AI by 2025. According to Jenkins, large language models that power chatbots are set to evolve into more efficient AI agents capable of handling straightforward tasks, thereby redefining user convenience and spearheading an era driven by AI-enabled ease in daily tasks. On the enterprise front, the rise of small language models (SLMs) is anticipated, with companies showing a preference for these models due to their ability to offer tailored insights without the reliance on high-end graphical processing units (GPUs). This trend is propelled by a heightened sensitivity towards data privacy, leading businesses to prefer localized deployments of SLMs that can be customized to their specific operational needs.
Advertisement
The term "intelligent agent" has emerged in the tech landscape as a significant concept, often described as a software or hardware entity capable of perceiving its environment and taking action to achieve designated goals. Intelligent agents exhibit autonomy and adaptability, deciphering changes in the environment through sensors or data inputs, and making informed decisions based on their learned knowledge and algorithms. This ability to independently learn and evolve is at the core of the intelligent agent's functionality, setting it apart from traditional chatbots, which typically respond to user prompts without context-aware decision-making capabilities.
The release of ChatGPT catalyzed a remarkable interest in generative AI, prompting numerous firms and investors to enter the AI sector. The competitive landscape, dubbed the "Hundred Model Wars," has witnessed a flood of announcements about new model upgrades from corporations like OpenAI, Anthropic, and Google, each vying for dominance in the AI space. However, challenges such as data scarcity and soaring hardware costs have hampered the speed at which large models can evolve, leading industry leaders to contemplate the sustainability of the current trajectory. Marc Benioff, CEO of Salesforce, highlighted this concern, suggesting that the excitement surrounding platforms like ChatGPT may have inflated expectations regarding AI capabilities.
Despite these setbacks, Jenkins remains optimistic about the future of intelligent agents and their potential applications. He cites that when ChatGPT surfaced in late 2022, it significantly shifted public perception of technology, prompting people to reconsider the possibilities of AI. Jenkins predicts that by 2025, intelligent agents will effectively assist individuals with tasks like scheduling appointments, online shopping, and managing bill payments—operating behind the scenes to minimize direct user interaction with screens. He envisions a future where users can expect a seamless experience, including the ability to book medical appointments without manual input, thereby allowing them to focus on more meaningful activities. This shift signals the dawn of an era characterized by convenience and efficiency, powered by artificial intelligence.
In keeping with these trends, companies across the globe have increasingly directed their attention towards developing intelligent agents. In China, AI firm Zhipu AI has recently upgraded its AutoGLM intelligent agent, while Lenovo launched its Lenovo Baiying intelligent agent aimed at small and medium-sized enterprises (SMEs). Meanwhile, Google Cloud is rolling out commercial AI agents to streamline client engagement processes, even offering a $300 credit for new customers to simplify their initial experience. NVIDIA has introduced the AI Blueprint to facilitate the development of visual AI agents which enhance operational efficiency through advanced video and image analysis. OpenAI has also announced plans to roll out "Operator," a product designed to execute complex tasks through an AI agent, slated for a release in January 2025, thereby broadening the landscape of AI agent applications.
As the industry moves toward a predominantly "small model" ecosystem, Jenkins pointed out that SLMs are set to gain substantial traction in enterprises by 2025. Their ability to provide customized insights while minimizing the dependency on powerful GPUs positions them as an ideal choice for businesses wishing to leverage large language models effectively. Importantly, SLMs can operate independently in local environments, reducing reliance on cloud computing resources and offering organizations greater control over their data privacy and compliance requirements—a crucial consideration amid rising concerns over data protection.
With ongoing developments, the modularity and scalability of SLMs will enable organizations to tailor these models to suit their distinct needs, thus allowing for adaptive responses to the rapidly evolving business landscape. Consequently, SLMs are not only fostering accessibility to AI but also serving as essential tools for navigating contemporary challenges associated with data management and privacy. Moreover, their lower deployment and maintenance costs make them appealing to budget-conscious SMEs, enabling them to harness AI capabilities previously available only to larger enterprises. The fast response times offered by SLMs promise to enhance operational efficiency, particularly in scenarios requiring immediate feedback. The inherent flexibility of these models facilitates their smooth integration into existing software ecosystems, resulting in seamless user experiences bolstered by advanced AI functionalities.
In conclusion, the AI landscape is on the brink of a transformative era, where small and intelligent agents will redefine our interaction with technology. As AI becomes an indispensable aspect of our daily lives and work environments, understanding these advancements will shape how we embrace this technology moving forward.