Ashlynn Hannah
January 29, 2024 | 6:20 pm
SHARE
Illustration of artificial intelligence adoption. (Pixabay/Tung Nguyen)
Jakarta. Indonesian companies are lagging behind their counterparts in Asia Pacific on generative artificial intelligence adoption, according to the PwC’s 27th Annual Global CEO Survey.
Around 53 percent of CEOs in Indonesia report generative AI has not yet been implemented in their companies, compared with 41 percent in the Asia Pacific region.
“In the past 12 months, half of the Indonesian CEOs (53 percent) report that their organizations have not yet implemented generative AI. But, over the upcoming year, about half of Indonesian CEOs expect it to enhance their ability to build trust with stakeholders (57 percent) and to improve product or service quality (56 percent),” said Eddy Rintis, PwC Indonesia Territory Senior Partner, in a statement on Monday.
According to Eddy, the survey found that seven out of ten CEOs believe AI in the next three years will bring increased competitiveness, changes to their business models, and the requirement of new skills from their employees.
However, 73 percent agree GenAI will increase cybersecurity risk, compared to 49 percent in Asia Pacific, and 53 percent agree it will spread misinformation, compared to 44 percent in Asia Pacific.
Despite threats and concerns, most CEOs globally, including 93 percent in Indonesia, have applied changes to their business models to keep up with the changing landscape. These changes are yet to have a positive impact on how CEOs see the future of their business, with 56 percent in Indonesia unsure what their company will look like in the next decade.
Most Indonesian CEOs believe that regulatory issues (75 percent) pose the biggest challenge in 2024, aligning with the Asia Pacific trend (66 percent). The regulatory environment takes the lead because CEOs feel less control over it compared to other challenges they can influence.
Following closely, technological capabilities (63 percent) rank as the second obstacle. Additionally, a lack of skilled labor and competing operational priorities are grouped as the third challenge, constituting 61 percent of the barriers.
PwC surveyed 4,702 global CEOs, including 1,774 in the Asia Pacific and 75 in Indonesia, from Oct. 2, 2023, through to Nov. 10, 2023. The global and regional figures are weighted proportionally to the country’s nominal GDP.
WHY ???
AI is often built around the English language, and poor knowledge of English can indeed hinder the adoption of AI in Indonesia. Here are some additional arguments to support this point:
Argument: AI systems are often trained on large datasets of text, which are typically written in English. This means that AI models are biased towards recognizing and understanding English language patterns, which can make it difficult for non-English speakers to interact with them effectively.
Sub-arguments:
- Language barriers: Many AI applications, such as chatbots and virtual assistants, are designed to understand and respond to English language inputs. Non-English speakers may struggle to communicate effectively with these systems, which can limit their ability to access AI-powered services.
- Data bias: The majority of data used to train AI models is in English, which means that the models may not be well-equipped to understand and process text in other languages. This can lead to biases and inaccuracies in AI decision-making processes.
- Limited access to AI resources: Many AI resources, such as online courses, tutorials, and research papers, are available only in English. Non-English speakers may struggle to access these resources, which can limit their ability to develop AI skills and stay up-to-date with the latest developments in the field.
- Difficulty in testing and debugging: Testing and debugging AI systems requires a deep understanding of the language in which they are designed. Non-English speakers may struggle to identify and fix errors in AI systems, which can lead to delays and inefficiencies in development and deployment.
- Limited availability of multilingual AI models: While there is an increasing trend towards developing multilingual AI models, many of these models are still limited in their language capabilities. Non-English speakers may struggle to find AI models that can understand and respond to their native language.
Conclusion: The dominance of the English language in AI development and deployment creates a significant barrier for non-English speakers who want to adapt and adopt AI technologies. To overcome this challenge, it is essential to develop more multilingual AI models and provide training and resources for non-English speakers to learn about AI.
