Ian Mulvany

January 22, 2022

Week 3 links of interest - more on AI, and a saxophone playing toilet!

Oh internet, you do have so so many things to read! Here are some of the things that piqued my interest in week 3 (and still so so many unread open browser tabs).

A nice twitter thread that builds towards a takedown of the strong Sapir whorf hypothesis - that language shaped our behaviour in fundamentally important social aspects.  The author's takeaway is that language brings us together far more than it divides us. 

2.  The guardian ran this piece on the possibility that machine learning could replace some of the main pillars of science - experiment and theory. When I came up through physics in the 90s we had already accepted that computation was a third pillar. I don't see anything new in this article, and the fact that many of the science references are the softer sciences just makes me overall not that impressed. What is more interesting is that the debate is getting an airing in the Guardian. 

3.  Hot on the heels of that piece, here is a nice piece on the limits of artificial intelligence, and indeed limits on how we can even make a determination on how smart ML systems are. The underlying conclusion perhaps leaves off a little unsatisfactorily, but I'd not heard about the Winograd schema challenge before. 

4.  I mentioned in an earlier post about the interesting rise of new kinds of research institutes. The Atlantic has a short piece about this: 

5.  The previous article refers to this article (https://www.pnas.org/content/118/41/e2021636118 ) by Johan Chu and James Evans. Anything James Evans publishes is worth following. The gist of the article is that as volumes of papers grow in a field, attention flows to more established papers  maybe because the competition for attention gets won in a winner takes all like scenario. I'll be honest, I didn't read the paper, but I used the excellent https://www.scholarcy.com to summarise the paper for me. The authors of the paper give some suggestions for how we might prevent this slowdown in innovation  like forced limits on papers that can be published by an author  but they recognise that many of the ideas they float are not that feasible. It's super fascinating to see hard evidence about the slowdown in progress, something that perhaps we are all worried about, but that it is hard to point to explicit examples of. 

The one thing I don't want to see as a response to work like this is the idea that what is needed are more "great men" with freedom to do "great things". This is basically the drum that Dominic Cummings has been banging on about since forever, and I think its horseshit.  


I had so many other things that I'd wanted to read this week, but next week is upon us already, and I'll keep combing for things of interest. I'll leave you with this though, an amazing episode from my favourite German children's TV series about a horse and sheep that live in a giant carrot. In this episode their toilet ends up saving a jazz gig by stepping in for the saxophonist. 

<iframe width="560" height="315" src="https://www.youtube.com/embed/m5AWkWtmtx0" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>