At the recent Meta Singapore’s inaugural Llama Incubator Program Demo Day, Minister for Digital Development and Information, Josephine Teo, addressed a crucial point on the perceived failure of AI experiments to deliver immediate returns for the global tech industry. While acknowledging external reports that “pour cold water on AI adoption,” Minister Teo insisted that such findings should not lead to pessimism but serve as a “reason to double down” on capability building and continued experimentation.
Minister Teo began her opening remarks by noting the significant progress Singapore has made, observing that AI, which “used to be on the fringe in Singapore, is increasingly in the mainstream”. This shift, seen everywhere from Singapore's CBD to community hackathons in the heartlands, is thanks to developers and strategic partnerships with global leaders like Meta.
Challenge of Technology Diffusion
A core theme of Minister Teo’s address was the need for AI to diffuse throughout the entire economy. Reflecting on her previous roles in the Manpower and Finance Ministry, she highlighted a key concern: “Unless technologies percolate into the corners of society, only a narrow group will benefit”. If AI remains concentrated only among frontier companies, the economy "does not realise its full potential". However, the diffusion of a general-purpose technology like AI is "easier said than done," requiring "many thoughtful interventions". The success stories showcased at the Llama Incubator Demo Day, a program established by Meta in partnership with the Singapore Government and various agencies like IMDA, GovTech, and EnterpriseSG, demonstrated this necessary mainstream adoption, offering sophisticated solutions to real-world issues. For instance, she cited MyRepublic’s AI Co-worker, which uses agentic systems to follow up on sales leads while "keeping the human in the loop, not replace them".
Learning from Setbacks
Addressing the skepticism surrounding AI adoption, Minister Teo acknowledged that reports noting that AI experimentation is failing are "fair comments, but they also reflect just how difficult experiments could be". She stressed that even if adoption is seen as a "fool's errand," valuable lessons must be learned. Minister Teo offered a historical analogy to underscore why the payoff for AI innovation takes time sharing a lesson in history “electrification was available from the late 1800s, but factories did not really use electricity until the 1920s.” This delay occurred because factories had to work around existing, embedded systems and continuously seek new breakthroughs. Teo believes the journey for AI adoption will be similar.
She emphasised that organisations need to find the “sweet spot” in experimentation to “build long term capabilities, and a sense of trust that the experimentations are serious and should not be given up on easily”. The crucial takeaway is that if experiments don't produce the desired immediate results, “we will get better at understanding the problem and build capabilities that enable the organisations to achieve success in AI in the future, we will still have gained something”.
Role of Open Source and Purposeful AI Incubators like the Llama Program are key enablers of this purposeful AI use. Minister Teo stated plainly that "Experimentation takes courage, but very importantly, it takes resources". The Llama program provides not only access to an advanced model but also the engineering resources, knowledge, and insight necessary for success.
Teo further noted that Singapore’s approach necessitates that the government "walk the talk when it comes to open source," citing the open-source nature of AI Verify, a testing framework and software toolkit. Looking at the practical use cases emerging from the program, Minister Teo categorised successful AI adoption into what she calls the "three P's and two A's":
Personalisation
Planning
Prediction
Automation
Anomaly detection (the ability for AI to search vast amounts of data for suspicious items)
Finally, Minister Teo urged participants not to "journey alone". Engaging with the community, such as groups like Lorong AI, allows practitioners to shorten their learning curve by leveraging the experiences of others, thereby solidifying the collective AI experience and keeping Singapore competitive.
