In a move that intensifies the global competition for technological supremacy, Alibaba has introduced a groundbreaking artificial intelligence (AI) model. This new model, designed to comprehend images and engage in more intricate conversations than its predecessors, marks a significant step forward for the Chinese tech giant.
According to a report from CNBC, Alibaba has announced two novel models, named Qwen-VL and Qwen-VL-Chat, both of which will be open source. This means that researchers, academics, and companies around the world can leverage these models to develop their own AI applications without the need to train entirely new systems, resulting in substantial time and cost savings.
The Qwen-VL model is capable of handling open-ended inquiries related to various images and can even generate descriptive captions for pictures. On the other hand, the Qwen-VL-Chat model is engineered for more intricate interactions. It can undertake tasks such as analyzing multiple image inputs and providing answers through several rounds of questioning. This includes creative tasks like composing stories and producing images based on user-provided photos, as well as solving visual mathematical equations.
One illustrative example provided by Alibaba involves an image of a hospital sign in the Chinese language. Using this image, the AI model can effectively respond to questions about the locations of specific hospital departments by interpreting the visual content.
Traditionally, the realm of generative AI—where technology produces responses based on human inputs—has primarily revolved around textual data. However, a recent advancement in this field is the incorporation of image understanding into AI models, mirroring the capabilities of Alibaba’s Qwen-VL-Chat and even the latest iteration of OpenAI’s ChatGPT.
Alibaba’s latest releases stem from its earlier Tongyi Qianwen large language model (LLM), which was introduced earlier in the year. An LLM, being trained on extensive datasets, forms the foundation for various chatbot applications.
This move follows Alibaba’s recent trend of open-sourcing AI models, with two other models having been made accessible this month. While this approach may not directly generate licensing revenue for Alibaba, the broader distribution of open-source AI models is anticipated to significantly expand the user base for these offerings. This strategic step comes at a crucial juncture when Alibaba’s cloud division is seeking to reinvigorate growth as part of its preparations for an impending public listing.
