ITB Professors’ Forum: Challenges and Potential of AI in the Future
BANDUNG, itb.ac.id — AI (Artificial Intelligence) technology has garnered the attention of many recently due to its benefits to people’s lives. However, its development still has its problems that need to be aware of.
To examine the development of AI further, the ITB Professors’ Forum held a discussion titled "Discriminative and Generative AI: Evolution, Opportunities, and Risks", with Prof. Dr. Ir. Bambang R. Trilaksono from ITB SEEI and Dr. Eng. Ayu Purwarianti, from ITB AI Center as its speakers on Saturday (27/5/2023).
Prof. Bambang explained that AI is not a new technology; it has been around since the 1950s as a simpler model. The models evolve as time passes, with increasingly complex functions and working mechanisms. Like the human brain, AI can perform problem-solving, perceive its environment, gain new information, and conduct reasoning on said information.
"From 2010 till now, humans developed deep learning, which is one aspect of machine learning where data can be extracted automatically," he said.
Deep learning uses a computational framework resembling an artificial neural network in humans. It is called an ANN (Artificial Neural Network). ANN can learn complex data patterns and generate representations based on specific algorithm functions. The two main types of paradigms in deep learning are discriminative and generative AI.
Discriminative AI classifies and predicts data based on the boundaries of the data’s class/category. On the other hand, generative AI can understand current data structures to generate new data.
"As one of the most popular AI nowadays, Chat GPT bases its system on a single generative AI model called a transformer. Its performance at some tasks has exceeded human capabilities," Prof. Bambang added.
Furthermore, Dr. Ayu added that AI will always inevitably bring destructive potential to people if not properly managed. The biggest risk of using AI is its direct or indirect impact on all facets of life including politics, social structures, economics, and science.
Therefore, AI management requires a framework with the foundation of six ethics: usefulness, safety and security, accountability, fairness, competence, as well as governance control.
"The European Union has already stated that the generative large language model, including Chat GPT, is classified as high risk on 24 February 2023. It is also Indonesia’s job eventually to decide our stance towards the development of AI soon,” Ayu stated.
Reporter: Hanifa Juliana (Urban and Regional Planning, 2020)
Translator: Ruth Nathania (Environmental Engineering, 2019)
scan for download
