147. Large language models (LLMs) - interview with Dr Guy Emerson

In order to understand learning, we need to understand the result of learning - expertise. This is much easier to approach in so-called "kind" domains, such as chess, where the rules are fixed and all information is available. However, there exist more "wicked" domains than this, such as tennis (where your opponent changes each match) or stock market investment (where the world is different each time). How do we study the development of expertise in fields such as these?

Chapter 22 of The Cambridge Handbook of Expertise and Expert Performance, entitled Toward Deliberate Practice in the Development of Entrepreneurial Expertise: The Anatomy of the Effectual Ask, concerns expertise in the art of entrepreneurship. This is a wicked domain par excellence, so much so as to throw into doubt the applicability or at least the generalisability of ideas about expertise from other domains, and yet the Handbook has a chapter approaching this topic, which is commendable. 

In this episode, you will hear about two key concepts that have arisen out of research on expert entrepreneurship - the Effectual vs. Predictive Frame; and the Entrepreneurial Ask. In other words, we will look at what research has to say about successful entrepreneurs' true attitudes vs. the popular conception in the media, and how they develop their skills.

Enjoy the episode.

***

RELATED EPISODES

125. Entrepreneurship education and conspicuous consumption

125+. Interview with Rasmus Koss Hartman

There has been a ton of research on how experts see things differently than novices. (Like, with their eyes.) Everything from where they look, how long they focus for, and their use of peripheral vision, to their ability to anticipate what is going to happen through picking up subtle visual patterns.

In this episode, I summarise and discuss this research.

Enjoy the episode.

Mindset was the first thing I spoke about on this podcast. I even did a separate episode going into the controversies surrounding replication of Carol Dweck's original work. Then there were stress mindsets, introduced by Kelly McGonigal in her book The Upside of Stress. (I happen to have also covered a book by her twin sister Jane, Reality is Broken, about applying the motivational principles learned by game designers in wider life situations).

But now I've encountered another kind of mindset: self-motivation mindset. Although the authors of Self-Regulation of Motivation: A Renewable Resource for Learning (2019) didn't name it that, it clearly is a type of mindset, in that it is a belief about oneself and one's potential. So now that we have not one, not two, but three mindsets to think about, I think it's time we tried to generalise as much as we can, and simply admit: mindsets matter. What other beliefs could there be that are affecting people's learning?

Enjoy the episode.

***

RELATED EPISODES

1. Mindset by Carol Dweck

68. The Upside of Stress by Kelly McGonigal

131. Mindset: Does it Replicate?

I haven't spoken on the podcast yet about my personal experience learning dancing. At university, I took part in dancesport, which is competitive ballroom and latin dancing; and in the last few years I have been learning to dance tango. I am struck by the differences in philosophies, skill sets, values, and learning cultures between these dance styles, so I wanted to share my experience with you.

Enjoy the episode.

***

Music used in this episode:

Uno by Anibal Troilo https://open.spotify.com/track/5TFzKLS8tjVMikVaOllr8L?si=69d0c8fbee934d2e

Orgullo Criollo by Osvaldo Pugliese https://open.spotify.com/track/74CjrywI50qOpaLrXo02ik?si=91b1644591a74407

Wake Me Up Before You Go-Go by Wham! https://open.spotify.com/track/0ikz6tENMONtK6qGkOrU3c?si=76304e73b1b04754

Dear Future Husband by Meghan Trainor https://open.spotify.com/track/3cU2wBxuV6nFiuf6PJZNlC?si=0dd4a5e2a23c46bc

...plus one surprise I won't spoil.

This is my first ever attempt at a VIDEO podcast. If you just listen to the audio, you should be fine.

This was a video produced for the STEM MAD conference in Melbourne in October 2023. Unfortunately I couldn't attend the conference, so I made this video to introduce the panel discussion on the role of generative AI in education.

Enjoy the episode.

This is a quick review of where I am now after 150 episodes and just short of 8 years of Education Bookcast.

Thanks for all of your support! Feel free to leave a review of the podcast, or, if you wish, support me on https://www.buymeacoffee.com/edubookcast .

Enjoy the episode.

Since I've now reached episode 150, I've decided to do something I've never done before - discuss a fiction book. (This episode contains spoilers.)

A Wizard of Earthsea is a fantasy novel from 1968, a time when the genre was still not very well-developed. Ursula Le Guin deliberately wanted to contravene some trends she saw in the existing genre, including the main characters being fair-skinned, and war as a moral analogy. In this book, the key issues are internal to a character, a fact that becomes increasingly clear as we read further.

The main character Ged (a.k.a. Sparrowhawk) goes through several educational regimes - a local witch who wants to take advantage of him; a regional wizard, Ogion, who hopes to provide him with the wisdom not to abuse his precocious powers; and a school, on the island of Roke, which teaches him all the knowledge he wants. Ged learns through bitter experience the value of Ogion's wisdom, though he spurns it as a child hungry for knowledge, power, and other people's approval.

I've read this book at least four times, and in three languages - English, Polish, and Spanish. Although its relevance to education is tenuous, I wanted to take advantage of episode 150 to talk about the book I've read the greatest number of times in my life.

Enjoy the episode.

A lot of the classic expertise research, especially the research about deliberate practice and the "10,000 hour rule", is inspired by K. Anders Ericcson's study of violinists at the Berlin Conservatory. However, we have seen before how misleading sampling a particular culture and generalising the findings over the whole of humanity can be. Thankfully, Lucy Green's How Popular Musicians Learn gives us something of an antidote to this classical music bias.

Green's book is based on interviews with 14 musicians in south-east England, of which 13 were instrumentalists and one, a singer. Their musical genres were all "guitar-based popular music" which includes rock and folk music. In her book, a number of findings undermine standard narratives about learning, including the inevitability of practice being unpleasant (the musicians enjoy their practice, unline classical musicains); the need for sheet music in order to learn (they all worked from recordings, and most couldn't read music); and the need for instruction (none of these musicians had been extensively formally trained, and those who had been had found it unhelpful).

Enjoy the episode.

***

RELATED EPISODES

Check out other episodes on anthropology and culture, and how they help provide wider samples for our understanding of psychology:

144. Developing Talent in Young People by Benjamin Bloom

121. Attachment Theory as Cultural Ideology

116. Cultural Foundations of Learning, East and West by Jin Li

106. The Anthroplogy of Childhood by David Lancy

89. The World Until Yesterday by Jared Diamond

39. The Geography of Thought by Richard NIsbett

SUPPORT

You can support Education Bookcast and join the community forum via Buy Me a Coffee using the following link: https://www.buymeacoffee.com/edubookcast .

Any teacher in a Western cultural context knows that classroom behaviour is the most challenging part of the job. A lot of the time, it seems like crowd control is the main issue, and "teaching" is secondary. Unfortunately, teacher training courses don't do a good job of preparing teachers for this reality, with behaviour management rarely instructed at all. 

Bill Rogers has been helping teachers develop their classroom behaviour management and discipline skills for decades. He has brought his calm and relationship-focused approach to innumerable schools, often including those with very challenging behaviour, or those in "special measures". His practical insights into what to do in the classroom, and the principles behind his approach, offer a valuable guide for teachers struggling with this aspect of their jobs.

I intend this to be one of several behaviour management books that I will cover on the podcast. Hopefully, in this way I can direct some teachers and school leaders to some useful resources, share some practical advice, and draw some general conclusions about school discipline and learning.

Enjoy the episode.

***

SUPPORT

You can support Education Bookcast and join the community forum via Buy Me a Coffee using the following link: https://www.buymeacoffee.com/edubookcast .

Dr Guy Emerson (a.k.a Guy Karavengleman) is a computational linguist working at the Cambridge University Computer Laboratory. In this episode, we discuss issues surrounding LLMs such as ChatGPT, GPT-3, GPT-4, and Google Bard.

Guy is concerned about misinterpretations of what the technology does and is capable of. As a computational linguist, he works on language models with a focus on semantics and human language acquisition, and thus questions of linguistic meaning and understanding are particularly relevant to his work. While LLMs are an impressive technology with startling capabilities, we need to be aware of when we may be fooling ourselves about their potential.

In this episode, we discuss what LLMs are; ways in which they have been misrepresented and misinterpreted; ethical questions about the companies developing this technology; and what they might be used for. 

Enjoy the episode.

***

SUPPORT

You can support the podcast and join the community forum by visiting https://www.buymeacoffee.com/edubookcast .