Indonesia wants to cease seeing synthetic intelligence (AI) merely as a know-how development and begin asking a extra elementary query: What nationwide problems does the nation need to remedy?
“Without that question, we will only become users, not creators,” mentioned Indonesia’s Vice Minister of Higher Education, Science and Technology, Stella Christie.
Christie delivered the keynote tackle at the AI Governance for the Greater Good: Balancing Innovation and Ethics occasion organised by the Centre for Strategic and International Studies (CSIS) in Jakarta on April 22.
According to Christie, having a transparent sense of objective would form the route of Indonesia’s future investments in AI, be it in expertise improvement or R&D round open-source ecosystems to align with nationwide priorities.
She warned that with no clear technique, Indonesia risked changing into extra like a knowledge provider for international AI corporations with out gaining added worth in return.
“I hope we could have more discussions about how to make AI useful for Indonesia, rather than the other way around,” she mentioned.
Rather than focusing solely on accelerating tech adoption or debating about the menace of automation, Christie urged the authorities, universities, and business to assume extra deeply about how Indonesia may construct significant AI capabilities whereas preserving uniquely human talents that machines can’t replicate.
Thinking processes mattered greater than outcomes
During her presentation, Christie highlighted the rising hype round generative AI (GenAI), significantly in the training sector.
In her view, AI literacy was typically misunderstood as merely understanding how to use AI purposes.
What mattered extra was the skill to assume critically and assess whether or not AI outputs had been correct and related, or doubtlessly deceptive.
She cited research from the Massachusetts Institute of Technology evaluating three teams of scholars: ChatGPT customers, search engine customers, and brain-only customers.
The examine discovered that college students who relied on their very own pondering talents produced the finest essays, whereas the ChatGPT group carried out the worst.
For Christie, the findings had been a reminder that training in the AI period shouldn’t focus solely on tech adoption, but in addition on serving to college students perceive the pondering processes behind adoption.
Ultimately, AI nonetheless didn’t be taught like people
Drawing on her cognitive scientist’s background when she used to examine how intelligence is constructed, Christie highlighted the distinction between how AI and the human mind be taught.
“One of the biggest differences between the human brain and AI is that humans can learn from very small amounts of data,” she mentioned.
Today’s AI methods stay closely dependent on huge datasets. In common, the bigger the dataset, the higher the mannequin performs.
Human intelligence, nonetheless, works otherwise.
A 3-year-old youngster may grasp their native language with out formal instruction.
Young kids additionally not often confuse on a regular basis objects comparable to cups or bicycles, regardless of encountering solely a restricted variety of examples all through their lives.
AI methods, against this, can nonetheless make errors even after being educated on hundreds of thousands of information factors.
“The human ability to understand concepts, form abstractions, and draw conclusions from limited information is an advantage we must preserve,” Christie added.
This was why the nation’s largest investments needs to be directed in the direction of expertise and human capital improvement, she emphasised, significantly in strategic areas related to Indonesia’s wants.
Ethics and innovation couldn’t be separated
Christie additionally rejected the concept that ethics and innovation had been inherently in battle.
She referred to the founding of Anthropic, the AI firm established by former OpenAI researchers over considerations about the route of AI improvement.
According to Christie, the resolution demonstrated that scientists had been nonetheless guided by integrity and ethical accountability.
She mentioned the message was significantly related for Indonesia, the place analysis and academia had been nonetheless typically seen as missing direct societal influence.
She pressured that scientific curiosity, not business demand, typically drives a few of the most transformation applied sciences that ship real-world worth somewhat than fast returns.
Indonesia between two international AI powers
During the Q&A session, Christie addressed considerations about the dominance of enormous AI fashions from the United States (US) and China over international locations in the Global South.
Responding to a query from a European Union delegate about bias and dependence on giant language fashions (LLMs), she mentioned Indonesia wanted to perceive its strategic place inside the international AI worth chain.
According to Christie, AI was essentially constructed on three primary elements: algorithms, information, and computing energy.
“When it comes to algorithms, we have to be realistic. It is extremely difficult for Indonesia to catch up with the US or China,” she mentioned.
The identical applies to computing infrastructure, which required investments of a whole bunch of hundreds of thousands of {dollars} simply to practice a single large-scale AI mannequin.
However, Christie believed Indonesia possessed one other essential asset, which is information.
As the world’s fourth most populous nation, Indonesia had an unlimited quantity of information that had but to be strategically utilised.
“If we control the data, we can buy algorithms and computing power,” she mentioned.
For that motive, she argued that international locations in the Global South wanted to begin treating information as a strategic asset somewhat than merely a by-product of digital exercise.
