Simon Chesterman 陈西文

Vice Provost & Dean @NUSingapore/@NUSCollege; @NUS_Law; Director @AISingapore | AI & data | international law | occasional fiction | the odd New Yorker cartoon

[Simon Chesterman 陈西文 @ProfChesterman]

I’m thrilled that my first work of general fiction, “Artifice”, is at last available worldwide! Skip the thread and order the book here: https://t.co/ZG2FZAIGbg
Or read on! #Artifice

[Simon Chesterman 陈西文 @ProfChesterman]

Technology should serve humanity. That proposition sounds simple. In the age of AI, it may be the most important article of faith we have.

[Simon Chesterman 陈西文 @ProfChesterman]

Pope Leo XIV’s new encyclical, 𝘔𝘢𝘨𝘯𝘪𝘧𝘪𝘤𝘢 𝘏𝘶𝘮𝘢𝘯𝘪𝘵𝘢𝘴, argues that artificial intelligence must serve humanity, not the other way around. Two groups now pore over its contents and implications.

[Simon Chesterman 陈西文 @ProfChesterman]

One is a mostly male, intensely hierarchical community built around revealed truth, specialised vocabularies, ritualised gatherings, and the promise that the faithful may one day transcend death.

The other is Catholic.

[Simon Chesterman 陈西文 @ProfChesterman]

Further reading https://t.co/TnlAZNkmrc

[Simon Chesterman 陈西文 @ProfChesterman]

That would be a real challenge for small and middle powers, where there is a strong desire to be at the table – rather than on the menu (H/T PM Mark Carney & Prof S Jayakumar).

[Simon Chesterman 陈西文 @ProfChesterman]

Yoshua’s answer was to learn more and to feel more – to embrace curiosity and not give up on the search for expertise, but also to develop our human qualities of connection and communication. No boos to that.

[Simon Chesterman 陈西文 @ProfChesterman]

Now that tariffs have been normalised, it’s not a big stretch to see how limiting access to models could serve economic nationalist goals as well.

[Simon Chesterman 陈西文 @ProfChesterman]

from a public that is growing increasingly sceptical of AI (though less so in Singapore and China than the US and Europe).

A fascinating development is the possibility that national security starts to limit access to the most powerful models.

[Simon Chesterman 陈西文 @ProfChesterman]

We’ve seen the rise and fall of export controls on chips, but on models themselves this first came up in respect of Anthropic’s Claude Mythos Preview as a cybersecurity concern.

[Simon Chesterman 陈西文 @ProfChesterman]

Successful responses need to show that responsible AI, governance, etc., operate not as stop signs but as seat belts – keeping us safe even as they enable us to drive faster. And it requires tangible benefits – and mitigation of harms –

[Simon Chesterman 陈西文 @ProfChesterman]

The key challenge is how to bring about change. It’s not always clear that fear is the best motivator; many engineers warn that AI will displace or destroy us all, then get back to work.

[Simon Chesterman 陈西文 @ProfChesterman]

A highlight was my chat with Yoshua Bengio on how concentrations of power risk encouraging our worst impulses in the development and deployment of AI – leading in turn to models that are duplicitous and manipulative (as they learned from the best: us).

[Simon Chesterman 陈西文 @ProfChesterman]

As always, an energizing and enriching week at ATx! Presentations and conversations ranged from opportunities in healthcare to aviation, from scientific discovery to public services – with a heavy dose of governance, which was front and centre in Minister Josephine Teo’s keynote.

[Simon Chesterman 陈西文 @ProfChesterman]

Colour me underwhelmed. The first great trial of the fastest moving technology in history ends because of a late filing. https://t.co/e9GDzmPA7Y

[Simon Chesterman 陈西文 @ProfChesterman]

I mean I'm all for encouraging class participation, but surely this goes a bit too far? https://t.co/5Hc2NMVHT4