The emergence of large-scale models has become a new trend in the AI industry. Many companies, including Microsoft, Aliyun, Baidu, and more, are investing heavily in developing large models to support various applications in offices, governments, e-commerce, finance, and travel industries. However, the shortage of chips has been a major obstacle to the practical application of these models.
To overcome this, companies are making efforts to attract top AI talent to develop large models. For instance, Meituan's co-founder Wang Huiwen said they would use 75% of the company's shares to attract top talents, while Li Kaifu is personally organizing Project AI 2.0. These measures show that companies are well aware of the importance of large models in AI development.
In March, Microsoft integrated GPT-4 into Office, making the office the first application that uses large models. Aliyun's "Tongyi Qianwen" and Baidu's "Wenxin Yiyuan" are other large model products developed for various industries, such as businesses, governments, and finance, to help users generate reports, PPTs, and more. The emergence of more and more large models means the market has more confidence in large models.
However, the development of large models is facing significant challenges due to the shortage of chips, similar to the situation faced by smartphones and electric vehicles. Training large models requires a lot of computing power, like building castles, and not everyone can use them. Moreover, the implementation of large models is also being delayed due to this issue. But many companies are now launching new large model projects amid these challenges. For example, after Baidu released "Wenxin Yiyuan," many companies launched their own large models almost simultaneously. These include Zhou Hongyi's "360 version of GPT," the launch of Aliyun's large language model "Tongyi Qianwen," and the announcement of SenseTime's large model "Xintianrun" and "Tiangong 3.5" jointly developed by Kungfutech and Qidian Zhiyuan. Zhihu and Mianbi Intelligence also cooperated to release a Chinese large model, "Zhihaitu AI." Furthermore, iFlytek also announced a "Spark Cognition Large Model" in May.
Without question, this year has become a boon for large models' development with the abundance of large model products becoming available. Overall, the AI industry is expanding rapidly, mainly due to the development of large models supported by top talent. The shortage of chips remains a significant obstacle that needs to be overcome, but companies are continuing to invest in this technology. The implementation of large models may be challenging, like digesting the stock market's behavior, but large models, especially when developed correctly, can indeed drive progress and lead the market to new heights.
