In the field of artificial intelligence, because deep learning algorithms are difficult to use in different application scenarios, some enterprises adopt the method of deep integration of technology research and development and application scenarios, and embed emerging technologies on the basis of existing technologies to achieve continuous innovation. Other companies focus on solving the problems of technology itself, are committed to developing a set of general technologies to replace existing products, and ignore the combination of technology research and development and application scenarios, that is, discontinuous innovation. Such different solutions lead to differences in the innovation model of enterprises.
In the process of transforming basic scientific research technology into commercial products, academic start-ups face the challenges of technological uncertainty and high risk. Such enterprises generally meet the following two conditions: first, at least one core technology has been transferred from the university to the newly established enterprise; Second, the founding member includes at least one inventor who is or has been affiliated with a university or research institute.
Since the technology around academic startups is still a certain distance from large-scale commercial application, it faces higher technological uncertainty and risks than other enterprises. At the same time, in the process of technology transfer from academia to industry, such companies also face legal constraints. Because in the process of technology commercialization, they need to cross from academia to industry, and there are differences in goals, values, norms and other aspects of the two fields, which will challenge the legitimacy of academic startups.
The uniqueness of academic startups increases their demand for complementary assets. To commercialize technology, academic startups need access to the necessary resources such as talent, capital, and connectivity, and crossing the boundary between academia and industry is a challenge that must be overcome. However, most of the founders of such enterprises are scientific researchers, who have been in an academic environment far from the market for a long time, lack in-depth understanding of user needs and industry and accumulated tacit knowledge, and their social network resources mainly come from academic resources such as schools, governments and large state-owned enterprises, and lack a strong network connection with the industry.
Therefore, access to external complementary assets to enhance enterprise capabilities is a key part of determining whether academic startups can create business value. This is different from other startups, where academic startups need to find niche markets in the early stages, but this is in stark contrast to entrepreneurial strategies that must acquire large amounts of complementary assets.
AI is a universal technology that empowers industries by reducing costs and improving efficiency (WIPO, 2019). Taddy (2019) states that AI represents a kind of system intelligence that absorbs human knowledge through computer vision and machine reading, digests, and uses this information to automate and accelerate the processing of tasks performed by humans. The artificial intelligence system mainly contains three elements: scene structure, data production and general machine learning, which reflect the large dependence of the artificial intelligence system on the knowledge structure and data in the scene it is applied.
For example, when applying AI technology to a segment, it is first necessary to structure the knowledge in the field and organize it into a specific structure so that the AI can identify. This step leads to a stretch forward in the product development process, making AI-based product development more complex than in the traditional information technology industry.
It can be seen that the high dependence of AI technology on application scenarios has further increased the problem of academic start-ups leaping from academia to industry. In the process of leapfrogging, some enterprises mainly focus on existing technologies and markets, and use emerging technologies to improve the performance of existing products or develop new product functions to achieve continuous innovation. This approach closely links emerging technologies to existing innovation ecosystems, promoting the integration of technology research and development with application scenarios.
However, some companies focus on the role of emerging technologies, aiming to disrupt the market and competitive landscape established by traditional technologies and achieve discontinuous innovation by creating entirely new products. This is especially common in academic startups. They rely on advanced technology in the laboratory and aim to replace existing technology and related products. However, product development from a technical perspective often faces the challenge of mismatch between technology development and application scenarios, which increases the difficulty of crossing the "valley of death".
