Jakarta (7/11/2019), Dessi Puji Lestari (Prosa.ai Chief Scientist of Speech) spoke at the ABDI DataGovAI Summit about the struggle of democratizing AI, specifically language processing, in Indonesia. The summit was held at JCC Senayan, Central Jakarta.
Dessi shared her experience of when she studied in Japan and how she witnessed the AI technology gap between Japan and Indonesia. Indonesia has adopted more AI technology than other ASEAN countries, according to the IDC report in 2018. But even so, Indonesia is still around 10 years behind compared to Japan. She highlighted that Indonesia actually has a big chance in AI, because not all technology can be adopted all at once in every country. The most obvious example is in the field of language processing.
Language processing is very dependant on the language. Text in Bahasa Indonesia can not be analyzed using an English analyzer. Words that are spoken in Bahasa Indonesia can not be automatically transcribed using an English Speech Recognizer. This language dependency issue can be solved comprehensively using local strength, both local engineers and local linguists that understand most about the culture and the language. Besides, Indonesia has more than 700 local languages, most of them are unofficial languages. The more diverse the language makes it more dependent and more contextualized. This means more effort needed by the local developer and local linguists because annotating these data requires a lot of language experts, time, and consistency.
These opportunities also come with challenges because the way people see AI is like a black box. People feel unsure when they don’t understand how the decision by the AI model was made. And because the adaptation of AI in Indonesia is still at an early stage, organizations are afraid to invest money in an AI-based project because it costs high thus does not have enough use cases in the market. Investors usually wait until other companies or competitors have succeeded with the technology before deciding on whether to invest in an AI project.
However, it is undeniable that AI will take place in our day-to-day life and Prosa.ai is trying to be part of this transformation by developing natural language processing and speech processing for Bahasa Indonesia. Language processing is shifting the way we interact with machine from keypads to a more natural and efficient way with speech-based interfaces. It automatically conducts many processes that we are now doing manually and clerically, such as analyzing people’s conversation and data analytics. It gathered meaningful insights that help us make better decisions.