Speak, an AI language learning platform officially launched in the Taiwanese market earlier this year and which has attracted a large number of users to download and use, and has also received investment from Open AI CEO Sam Altman and others, today (October 10st) co-founder and CEO Connor Zwick and co-founder Andrew Hsu came to Taiwan to share their founding philosophy and future market layout plans.
Simulating real-life scenarios allows users to learn English more intuitively
When Speak was founded, we saw the global market demand for English learning and believed that English plays an important role in many forms of communication, interaction, learning and understanding today. In addition, due to our own experience in learning foreign languages, we are aware of the challenges faced by non-native speakers when learning a new language, as well as the problems in the current learning arrangements of many language courses.
Connor Zwick said that many languages can actually be learned more efficiently through daily life than through school, cram schools, tutoring, and other channels. The reason is that English is used during the learning process, but in daily life, people are still accustomed to using their native language. In addition, the content of many courses is relatively standardized, so it is difficult to use words and sentences similar to those used by ordinary native speakers, which makes the process of learning English more "difficult."
Speak's initial focus is on speech recognition technology, enabling the system to more clearly identify English sentences spoken in different accents and non-native ways, thereby determining whether the user's pronunciation or statement are incorrect and providing assistance with corrections. Furthermore, through artificial intelligence technology, it simulates different scenarios and roles, making it easier for users to learn how to use English in various interactive situations and making it easier for users to learn English through conversation.
Building a learning platform through three major public clouds and artificial intelligence technology
In current technology applications, Speak distributes computing loads through Google Cloud and AWS cloud service platforms. It can not only obtain greater computing power through the cloud platform, but also distribute access traffic through different cloud service platforms, thereby maintaining stable service operations and faster response efficiency.
As for the application of artificial intelligence technology, Connor Zwick emphasized that in addition to building artificial intelligence models on its own for more than 7 years, it also began to cooperate with OpenAI in 2022 to use its GPT-4 and above artificial intelligence model computing capabilities to correspond to more natural interactions and dialogue application functions that can simulate roles in different scenarios.
In addition, the current cooperation with OpenAI is mainly through API resource connection, and behind these resources is also the Microsoft Azure cloud service platform. Therefore, the overall resources of the three major public cloud service platforms are integrated to meet different computing needs.
The Taiwanese market became Speak's key focus at the beginning of this year.
Taiwan is currently Speak's market expansion after South Korea and Japan. Connor Zwick said that when planning to enter the South Korean market, they had actually noticed that the Taiwanese market had a significant demand for English learning. Therefore, the Taiwanese market became Speak's key layout location at the beginning of this year, and it has accumulated significant growth performance. It has also launched elementary courses, travel English and workplace English specifically for the Taiwanese market to meet the English learning needs of more Taiwanese users from an early age, as well as those when traveling abroad and looking for jobs.
In addition, Connor Zwick also took the Taiwanese market as an example. Although both use Chinese as the main language of communication, the traditional Chinese characters commonly used in Taiwan are indeed different from the Chinese usage in other places such as China. Therefore, localized fine-tuning will be carried out in artificial intelligence models and data applications to better meet the localized learning interface interactive experience and make it easier for Taiwanese users to learn English.
Adhering to the Speak team's elite recruitment system (currently only 75 people), the Taiwan team currently has only two employees responsible for local promotion, but it is expected that they will continue to recruit talents from outside and hope to continue to listen to feedback to create more features that meet local English learning needs.
We will expand to more markets and provide more language learning services in the future.
As for whether there are plans to expand into China, a country with a larger learning market, Connor Zwick stated that they do have such plans, but they will undoubtedly utilize local AI models and related service technologies. Regarding expansion into other markets like Europe, which have more stringent regulations regarding the use of AI services and access to personal data, Connor Zwick also stated that they will invest in accordance with local regulations.
Regarding the current use of artificial intelligence to enable more people to learn English efficiently, whether the same method will be considered in the future to assist in learning different languages, Connor Zwick said that there is indeed such a plan. In fact, a Spanish learning service has already been launched, and it is expected that more language learning will be added in the future. At the same time, the existing learning services will be expanded to support more languages, allowing more native speakers to easily learn English.
