How close is AI to taking my job? (And what the benchmarks aren't telling us)
Are we there yet? From viral benchmarks to billion-dollar tests, AI researchers are racing to predict when automation will reshape our careers. But according to Anson Ho, we might be looking in all the wrong places.
Join AI Safety Hong Kong for the second instalment of our Speaker Series, featuring Anson Ho, a researcher and writer at Epoch AI. Anson’s work focuses on understanding the trajectory of AI developments, from the economics of transformative AI and algorithmic progress to US-China AI competition and AI benchmarks.
In this talk, Anson will tackle one of the most urgent questions of our time: "How close is AI to taking my job?"
Drawing from his popular deep dive on the Epoch AI substack, he will argue that while flashy benchmarks like OpenAI's GDPval suggest AI has already surpassed human experts, the real-world economic impacts remain surprisingly muted. Why the disconnect? Because we have been “searching under the streetlight”—tracking what is easy to measure rather than what matters.
Anson will share his unique approach to cutting through the hype: actively trying to get today's best AI agents to do his actual job. He will walk us through three real-world tasks (coding, writing, and publishing articles), revealing where AI excels, where it fails in spectacularly human-like ways, and crucially, when he expects these bottlenecks to finally fall.
Key discussion points will include:
Why million-dollar benchmarks are failing as leading indicators for job automation, and what a more honest, real-world approach to measuring job automation actually looks like
What happens when you try to get AI to do your own job, and what that reveals about where current models still fall short
Realistic timelines for AI automation in coding, writing, and administrative tasks from 2026 to 2029.
What the automation of knowledge work means for Hong Kong — and whether its status as a global financial hub is a strength or a vulnerability in the age of AGI
Who should attend: This talk is for anyone who has ever wondered if their job will be the next to go. AI researchers, software engineers, writers, economists, policy professionals, students, and all curious minds are welcome. No technical background is required.
About the Speaker:
Anson Ho is a researcher and writer at Epoch AI. His prior work covers the economics of transformative AI, quantifying the rate of algorithmic progress, breaking down the dynamics of US-China AI competition, and developing AI benchmarks. He is interested in better understanding the trajectory of AI developments to help inform society, including AI researchers and policymakers.
Stay connected with AI Safety Hong Kong.
LinkedIn: https://www.linkedin.com/company/ai-safety-hong-kong/
Substack: http://substack.com/@aisafetyhk
Website: https://www.aisafetyhk.org/