Financial giantsGoldman SachsIn an interview looking ahead to 2024, Marco Argenti, chief investment officer of , said that the year will focus mainly on "mixed"AIand built onLarge Language ModelsApplications above.
He stressed that "hybrid AI" will become a dominant trend, in which large models act as the "brain" that interprets prompts and user needs, or as an orchestrator that plans tasks by assigning tasks to multiple working models specialized for specific tasks, similar to ChatGPT.
Source Note: The image is generated by AI, and the image is authorized by Midjourney
However, Argenti also pointed out that constructing similarGPT-4Large programs like OpenAI’s will be too expensive for only the wealthiest companies like OpenAI to afford. So he expects most companies will focus on building smaller neural networks, either in their own data centers or on cloud computing services, that are trained on their proprietary data.
In addition to the discussion of hybrid structures, Argenti also foresees a class of third-party applications built on top of the underlying models coming in 2024. He likened these underlying models to operating systems or platforms, and believes that a whole set of applications that have not yet appeared around these models will emerge in the near future. He said this creates a huge opportunity for capital to shift to the application layer and tool layer, and this shift may happen as early as next year.
Despite his positive outlook on the new trend, Argenti also stressed the importance of coordinating security among different parties. He stressed that in order to promote the development of technology, it will be necessary to continue to foster an environment that encourages collaboration, encourages open source models when appropriate, and develops appropriate principle-based rules to help manage potential risks. These risks include bias, discrimination, security and robustness, and privacy.
Argenti is confident that by 2024, companies will focus on showingHighestHe believes that with the rise of applications, capital will be more inclined to flow to the application layer and tool layer, promoting further development in the field of artificial intelligence.