• ICMI 2025 Keynote – Future of Human Oversight

    ICMI 2025 Keynote – Future of Human Oversight

    It was a great pleasure to deliver ๐—ฎ ๐—ธ๐—ฒ๐˜†๐—ป๐—ผ๐˜๐—ฒ ๐˜†๐—ฒ๐˜€๐˜๐—ฒ๐—ฟ๐—ฑ๐—ฎ๐˜† ๐—ฎ๐˜ ๐˜๐—ต๐—ฒ ๐Ÿฎ๐Ÿณ๐˜๐—ต ๐—”๐—–๐—  ๐—œ๐—ป๐˜๐—ฒ๐—ฟ๐—ป๐—ฎ๐˜๐—ถ๐—ผ๐—ป๐—ฎ๐—น ๐—–๐—ผ๐—ป๐—ณ๐—ฒ๐—ฟ๐—ฒ๐—ป๐—ฐ๐—ฒ ๐—ผ๐—ป ๐— ๐˜‚๐—น๐˜๐—ถ๐—บ๐—ผ๐—ฑ๐—ฎ๐—น ๐—œ๐—ป๐˜๐—ฒ๐—ฟ๐—ฎ๐—ฐ๐˜๐—ถ๐—ผ๐—ป, ๐˜„๐—ต๐—ฒ๐—ฟ๐—ฒ ๐—œ ๐—ฒ๐˜…๐—ฝ๐—น๐—ผ๐—ฟ๐—ฒ๐—ฑ ๐˜๐—ต๐—ฒ ๐—ฒ๐˜ƒ๐—ผ๐—น๐˜ƒ๐—ถ๐—ป๐—ด ๐—ป๐—ฎ๐˜๐˜‚๐—ฟ๐—ฒ ๐—ฎ๐—ป๐—ฑ ๐—ณ๐˜‚๐˜๐˜‚๐—ฟ๐—ฒ ๐—ผ๐—ณ ๐—ต๐˜‚๐—บ๐—ฎ๐—ป ๐—ผ๐˜ƒ๐—ฒ๐—ฟ๐˜€๐—ถ๐—ด๐—ต๐˜. Recent reports such as OpenAIโ€™s GDPVal and METR show that AI systems can now autonomously perform complex doing/solving tasks and often surpass human experts on…

    Read more: ICMI 2025 Keynote – Future of Human Oversight
  • Representing Australia in the Global Effort for AI Safety

    Representing Australia in the Global Effort for AI Safety

    It has been a great honour to represent Australia on the Expert Advisory Panel for the International AI Safety Report, led by Joshua Bengio and backed by over 30 countries and international organisations including the EU, OECD and UN. This is the most important ongoing effort to monitor AI capability gains and safety risks of…

    Read more: Representing Australia in the Global Effort for AI Safety
  • When โ€œRisk-Basedโ€ AI Becomes an Empty Promise

    When โ€œRisk-Basedโ€ AI Becomes an Empty Promise

    ๐—ฅ๐—ถ๐˜€๐—ธ-๐—ฏ๐—ฎ๐˜€๐—ฒ๐—ฑ ๐—”๐—œ ๐—ฝ๐—ผ๐—น๐—ถ๐—ฐ๐˜†/๐—ฟ๐—ฒ๐—ด๐˜‚๐—น๐—ฎ๐˜๐—ถ๐—ผ๐—ป ๐˜€๐—ผ๐˜‚๐—ป๐—ฑ๐˜€ ๐—ผ๐—ฏ๐˜ƒ๐—ถ๐—ผ๐˜‚๐˜€. ๐—•๐˜‚๐˜ ๐—ถ๐˜ ๐—ฟ๐—ฒ๐˜€๐˜๐˜€ ๐—ฎ ๐—ณ๐—ฟ๐—ฎ๐—ด๐—ถ๐—น๐—ฒ ๐—ฎ๐˜€๐˜€๐˜‚๐—บ๐—ฝ๐˜๐—ถ๐—ผ๐—ป: ๐˜๐—ต๐—ฎ๐˜ ๐˜„๐—ฒ ๐—ฐ๐—ฎ๐—ป ๐—ฎ๐—ฐ๐˜๐˜‚๐—ฎ๐—น๐—น๐˜† ๐—บ๐—ฒ๐—ฎ๐˜€๐˜‚๐—ฟ๐—ฒ ๐—ฟ๐—ถ๐˜€๐—ธ.Yet risk assessment is often put forward as the first thing to answer, as if itโ€™s the easy part. Some frameworks push โ€œuse caseโ€“basedโ€ solutions with a pre-defined list of high-risk use cases, but then smuggle in the magic…

    Read more: When โ€œRisk-Basedโ€ AI Becomes an Empty Promise
  • From Shadow Adoption to Workflow Collapse: The Hidden Economics of AI

    From Shadow Adoption to Workflow Collapse: The Hidden Economics of AI

    ๐—ช๐—ต๐—ฎ๐˜ ๐—ถ๐—ณ ๐˜„๐—ฒโ€™๐˜ƒ๐—ฒ ๐—ฏ๐—ฒ๐—ฒ๐—ป ๐—ฐ๐—ฎ๐˜๐—ฒ๐—ด๐—ผ๐—ฟ๐—ถ๐˜€๐—ถ๐—ป๐—ด ๐—š๐—ฒ๐—ป๐—ฒ๐—ฟ๐—ฎ๐—น ๐—ฃ๐˜‚๐—ฟ๐—ฝ๐—ผ๐˜€๐—ฒ ๐—”๐—œ (๐—š๐—ฃ๐—”๐—œ) ๐˜๐—ต๐—ฒ ๐˜„๐—ฟ๐—ผ๐—ป๐—ด ๐˜„๐—ฎ๐˜†?โ€ข What if it isnโ€™t just another โ€œgeneral-purpose technologyโ€ like computing, but a bundle of many specific-purpose technologies out-of-the-box?โ€ข And what if calling AI a โ€œcapital investmentโ€ is also a category errorโ€”when in reality it behaves more like inexpensive labour? Yesterday at the Cost-Benefit…

    Read more: From Shadow Adoption to Workflow Collapse: The Hidden Economics of AI
  • If You Understand It, Itโ€™s Not AI: Designing Oversight for the Incomprehensible

    If You Understand It, Itโ€™s Not AI: Designing Oversight for the Incomprehensible

    New paper alert (working draft)! โ€œWhen you really understand what AI is doing, itโ€™s no longer AI โ€” itโ€™s just boring automation.โ€ That old AI community joke captures the paradox of human oversight. We trust calculators and complex business process engines with millions of ifโ€“else rules, running hundreds of steps, responding to environmental changes, and…

    Read more: If You Understand It, Itโ€™s Not AI: Designing Oversight for the Incomprehensible
  • Writing in the Age of AI
    ,

    Writing in the Age of AI

    After my slightly controversial talk on ๐—ฅ๐—ฒ๐—ฎ๐—ฑ๐—ถ๐—ป๐—ด ๐—ถ๐—ป ๐˜๐—ต๐—ฒ ๐—”๐—ด๐—ฒ ๐—ผ๐—ณ ๐—”๐—œ, I delivered another talk at APS Learn series on ๐—ช๐—ฟ๐—ถ๐˜๐—ถ๐—ป๐—ด ๐—ถ๐—ป ๐˜๐—ต๐—ฒ ๐—”๐—ด๐—ฒ ๐—ผ๐—ณ ๐—”๐—œ. ๐— ๐˜† ๐—ฐ๐—ผ๐—ฟ๐—ฒ ๐˜๐—ต๐—ฒ๐˜€๐—ถ๐˜€: ๐˜„๐—ฟ๐—ถ๐˜๐—ถ๐—ป๐—ด ๐—ถ๐˜€ ๐˜๐—ต๐—ถ๐—ป๐—ธ๐—ถ๐—ป๐—ด. We write first for self-understanding, then for audience understanding. AI cannot replace that understanding, but it can be a strong discussion partner.…

    Read more: Writing in the Age of AI
  • The AI-ESG Paradox: Why Assessing AI’s Impact Defies Simple Metrics

    The AI-ESG Paradox: Why Assessing AI’s Impact Defies Simple Metrics

    Just sat down with the latest episode of The Institutional Edge: Real allocators. Real alpha. โ€” and this one hits right at the crossroads of technology, investing, and sustainability. Angelo Calvello speaks with me about why AIโ€™s ESG impact is so hard to measure and what investors can realistically do about it. We dig into…

    Read more: The AI-ESG Paradox: Why Assessing AI’s Impact Defies Simple Metrics
  • Zoho Fireside Chat on AI

    Zoho Fireside Chat on AI

    I never imagined I would one day share a speaking session โ€” and a speaker room chat โ€” with the ๐—น๐—ฒ๐—ด๐—ฒ๐—ป๐—ฑ๐—ฎ๐—ฟ๐˜† ๐—”๐˜‚๐˜€๐˜๐—ฟ๐—ฎ๐—น๐—ถ๐—ฎ๐—ป ๐—ฐ๐—ฟ๐—ถ๐—ฐ๐—ธ๐—ฒ๐˜ ๐—ฐ๐—ฎ๐—ฝ๐˜๐—ฎ๐—ถ๐—ป ๐—ฆ๐˜๐—ฒ๐˜ƒ๐—ฒ ๐—ช๐—ฎ๐˜‚๐—ด๐—ต. Thanks to the Zoho event in Sydney, that happened. Speaking right after Hugh Watson Australia’s Ambassador for Cyber Affairs and Critical Technology also added a valuable global perspective to…

    Read more: Zoho Fireside Chat on AI
  • Panels at WA Gov Innovation Week

    Panels at WA Gov Innovation Week

    I was glad to join the ๐—ช๐—ฒ๐˜€๐˜๐—ฒ๐—ฟ๐—ป ๐—”๐˜‚๐˜€๐˜๐—ฟ๐—ฎ๐—น๐—ถ๐—ฎ๐—ป ๐—š๐—ผ๐˜ƒ๐—ฒ๐—ฟ๐—ป๐—บ๐—ฒ๐—ป๐˜โ€™๐˜€ ๐—œ๐—ป๐—ป๐—ผ๐˜ƒ๐—ฎ๐˜๐—ถ๐—ผ๐—ป ๐—ช๐—ฒ๐—ฒ๐—ธ, contributing to discussions on both AI and cyber. The event was buzzing with energy, sharp questions, and concrete use cases being shared. On cybersecurity, my key messages were:1. You donโ€™t need to copy-share data everywhereโ€”doing so only increases the attack surface. Analytics and compute…

    Read more: Panels at WA Gov Innovation Week
  • Future-Proof Protection for Sensitive Data

    Future-Proof Protection for Sensitive Data

    CSIRO’s Data61 latest research gives copyright owners and data producers a new superpower โ€” the option to make their sensitive and sovereign data ๐˜‚๐—ป๐—น๐—ฒ๐—ฎ๐—ฟ๐—ป๐—ฎ๐—ฏ๐—น๐—ฒ by AI. Itโ€™s been getting some attention, with us appearing on ABC TV, 7 News, and 2GB radio to explain the technology. Unlike past โ€œconfuse the AIโ€ tricks like data poisoning,…

    Read more: Future-Proof Protection for Sensitive Data

Authored Books

About Me


About me – According to AI

Director/Head of CSIRO’s Data61
Conjoint Professor, CSE UNSW

For other roles, see LinkedIn & Professional activities.

If you’d like to invite me to give a talk, please see here & email liming.zhu@data61.csiro.au

Featured Posts