The history of computing took a turn when we started to study the interaction between people and computers from the perspective of humans as psychological and sociological beings rather than components of a machine. With each new era of technology, this perspective has proven helpful, here we apply it again to AI, and the current global interest in large language models.

To address this interdisciplinary perspective we will overlap with the Psychology of Programming Interest Group, a 30+ year group who have studied human perspectives on computer programming. We will host four hours of talks, discussions and invited talks.

When: 23 August 2023 at 9.30 to 15.30 CET

Where: Teknodromen, M-building, Ole Römers väg 1, Lund, Sweden and online

Spoken language: English

Programme outline

09.30 Fika & registration

09.55 Online meeting opens

10.00 Morning session - also online

Large Language Models and the Psychology of Programming

Invited speaker: Professor Clayton Lewis
College of Media, Communication and Information, University of Colorado Boulder, USA

Abstract: Large Language Models have emerged rapidly as powerful coding tools, in some cases showing the ability to create entire working programs, and (more commonly) providing help with the details of a great many APIs and frameworks. This emergence raises a number of questions for the PPIG community. Will these tools change what programmers do, in ways that affect "the psychology of programming"? Given the (apparent) command that these systems have of natural language, and of the semantics of a great many domains of activity, can they be used to enhance the kinds of interactions with software tools that PPIG researchers have studied? Noting that large language models exhibit analogical reasoning as an emergent capability, can they leverage insights from early research on programming by analogy? Does predictive modeling, as a key cognitive process that can be applied in many domains, suggest new ways to think about programming not based on text?

Interactive session

We will then have an interactive session where members of the Psychology Programming Interest Group and members of the AI Lund network will collaborate together using LLM-based technology to build something creative! No coding experience needed.

12.15 Lunch

13.15 Afternoon session - also online

After lunch there will be academic presentations contextualising recent advances in AI in the history of programming and exploring research questions in how we can design user interfaces to allow everyone to interact, control and manage new forms of AI technology.

Presentations

• Interactive Narrative Visualization for Learning Markov Decision Process
  B. Mbuu Mutua and Alan F. Blackwell
• Participatory prompting: a user-centric research method for eliciting AI assistance opportunities in knowledge workflows
  Advait Sarkar, Ian Drosos, Rob Deline, Andrew D. Gordon, Carina Negreanu, Sean Rintel, Jack Williams and Benjamin Zorn
• Prompt Programming for Large Language Models via Mixed Initiative Interaction in a GUI
  Tanya Morris and Alan Blackwell
• Back to the future: What do historical perspectives on programming language research tell us about LLMs?
  Tao Dong and Luke Church

15.30 Fika and mingel

We'll conclude with a mixing session where the two communities can continue the conversation.

---

Organisation

• Söderberg Emma, Senior Lecturer, Department of Computer Science, Lund University
• Luke Church, Adjunct Lecturer, Department of Computer Science, Lund University
• Jonas Wisbrant, AI Lund, Lund University