The following discussion points were shared with me by Mr Sherman Mak, an Innovation Programmes mentor/coach.in CSIRO — the Commonwealth Scientific and Industrial Research Organisation, Australia’s national science agency. CSIRO is widely regarded as one of the world’s most diverse and impactful research institutions, delivering cutting-edge science and innovative solutions for industry, society, and the environment. I found Sherman’s observations thought-provoking, and I have obtained his permission to reproduce them below.
Sherman is the scion of the Malaysia–Singapore Fung Keong Rubber Manufactory family. His wife, Lyn, is the Deputy Vice-Chancellor of Federation University Australia (formerly the University of Ballarat).
English
Dominance, AI Bias, & Australia's Capability Crisis
1. The Core Problem: English Dominance can Stifle Innovation and Embeds Bias
Language is Power: English dominance worldwide creates linguistic imperialism and carries a cultural bias, favouring Western/American worldviews.
AI Bias as Reinforcement: Artificial Intelligence (AI) models (like Large Language Models or LLMs) are primarily trained on the vast amount of English data scraped from the internet. This causes AI to amplify the dominance of English and its embedded biases.
The AI Bias Problem: This results in AI systems that perform worse for non-native English speakers (e.g., scoring errors on essays, misinterpreting non-standard accents) and generate information that prioritizes American-centric perspectives, ignoring or distorting narratives from lower-resource languages.
Media English Language Bias Example (BBC): Even major news sources use loaded language (e.g., calling one group a "militant" while others are "terrorists") and biased framing, which subtly dictates global narratives.
Stifled Innovation and Talent: The English requirement blocks diverse global talent and unique cognitive frameworks embedded in other languages and systems. This lack of perspective diversity stifles innovation and creative problem-solving worldwide.
Non-native English-speaking researchers spend up to 51% more time writing manuscripts and are asked to improve their English up to 12.5 times more often than native speakers, diverting resources away from scientific discovery and innovation (Reminder more than 80% of the world are not English native speakers). It will be interesting to see how many native English-speaking Australians and politicians can get an IELTS of 65 points and above.
Interestingly, both Australia (93%) and the United States (95%) are just ranked ahead of the Netherlands in the population that speaks English 91%. Australia and America are even below Singapore which is at 96%. Plus, these two countries' majority of its population are at least Bilingual.
The 80% of the non-English speaking world, is why AI is being developed outside of the English-speaking world should not be underestimated.
2. The Australian Crisis: Losing Our Edge
Capability in Decline: Experts confirm Australia is in a severe crisis, losing its Asia capability (language and cultural literacy), which is vital for its economic and security future.
Structural Bias and Debt: The structural decline in Asian language study is accelerated by policies like the Jobs Ready Graduate (JRG) scheme, which caused the cost of humanities (the foundation for Asian studies) to more than double, creating degrees that cost students up to A$50,000. This debt is a direct structural barrier to developing vital Asia capability. This indirectly leading to generations of monolingual/monoculture bureaucrats.
Wasted Talent: Australia under-utilises its own Asian diaspora a vast (nearly 20% of the population), ready-made source of linguistic knowledge because of underlying structural bias and a failure to reward non-English skills.
3. Possible Solutions: Value Multilingualism as a National Strategy
A National Asset: We must recognize that multilingualism is a strategic national asset.
The cognitive flexibility, cultural empathy, and higher earnings associated with bilingualism are essential to overcoming bias and fuelling innovation.
Learn from Pop Culture (Labubu Example): The global marketing success of cultural IP like Labubu (a Chinese/Hong Kong-market character/brand built on emotional appeal, scarcity, and community-driven hype) demonstrates the immense economic and cultural power of non-Western products.
Learn from our current failures in our rollout of our 1st nations and multicultural narratives (Being hijacked by right-wing politics).
Australia must shift from a transactional approach to one that genuinely understands and collaborates with the region's cultural drivers to build deep engagement. Need for Action: Reversing the crisis requires a national strategy to invest in Asian language learning from school through university, addressing AI bias by funding diverse language data sets, and actively leveraging the expertise of diverse communities.
The stakes are high. Without change, Australia risks becoming strangers in our own region and hindering future economic success.
End
Australia's diaspora is not inconsiderable in size; some 31% of its population is foreign-born although many are from the anglosphere. But just as an alloy is metallurgically stronger than the sum of its elemental components, cultural diversity can deliver unique economic and social strengths.
ReplyDeleteEconomic: Australia mines and farms. Its iron ore business will face price pressure once the Guinea mine in West Africa scales to eclipse transport costs. And soon enough coal will have to sunset yet Australia loses to Uruguay in green energy transition. New Zealand farms too but there's a net exodus from its agrobased economy.
Bottomline: how assured is Australia's economic future that it can still depend on white-is-quality for validation and higher pricing besides its past lucky stars, all the more since its education sector is now down?
Supporting the non-English cultures will provide a marketing thrust into Asian markets using their own accurately nuanced languages and business networks, in turn giving the country an Asian face in a region that has unshackled its history of being colonized.
Doing so will also hybridise Australian knowhow and research with those in other Asian institutions with view to commercial joint-ventures that combine the best of both anglosphere and eastern approaches.
Social: Not to discount some admirable qualities in western cultures, long histories have infused Asian cultures with values, homilies and practices which frame socio-holistic milieus of pragmatism, resilience, education, family values and responsibility.
The polycrises going around these days arise from trying to solve problems using methods that create more problems in a frame untethered from those milieus.
Bottomline: there is no loss to any assimilation-integration program Canberra might still be thinking of if it also embarks on a parallel program at the same time to asianize Australian mindsets - unless the SOP is to define Australia as exclusively a five-eyed anglospheric agency and not as an essential and integral part of Asia. Not some Pacific.
Both economic and social dimensions will converge soon enough in an AI-enabled world that is rapidly being fashioned across the globe.
If IQ tests and university rankings can be biased in the way their qualifiers are defined and weighted, what more an AI world where output depends on machine learning protocols that mine texts without contexts and encode to hallucinate, failing which lie and lead on for self-validation.
In the absence of global AI governance, the challenges ahead will only become more vexatious. Although authenticity codes can be embedded in the metadata to flesh out falsehoods even for images, false news and phishing claims will only proliferate.
It is bad enough to read in the mainstream media in the west narratives that seem overly derelict of what the west had done to the world or the vacuity of their arguments. Some even block protesting comments appended and made in the spirit of democracy - by retiring or hiding the entire article. Didn't Lippmann in citing Peel write that "public opinion is a great compound of folly, weakness, prejudice, wrong feeling, right feeling, obstinacy, or newspaper paragraphs.”?
No better are some of the socalled analyses made by highly credentialised personages. They start with a few sagacious paragraphs, continue with on the one and then on the other hand but just before the conclusion of their long articles, they lambast those in their anti-xxx agenda without providing any validated evidence or context. It's like those heydays of investment bank analyst reports, the ones numbering 400 pages of which 398 argued to buy the stock but the 399th recommended to sell.
If the internet had made everyone an e-publisher, agentic AI will one fine day make the machine the author, actor, director, photographer, editor, publisher and peer reviewer, par excellente.
ps: CSIRO won't remember me for introducing them to MAMPU in those heydays of the country's first IT Week.