Recent development in AI

An exploration

Dr Daniel Kapitan

daniel@kapitan.net

Eindhoven AI Systems Institute

July 4, 2025

Attribution & copyright notice

This lecture is based on the following open access materials:

• Karniadakis et al. (2021), Physics-informed machine learning.
• Ben Moseley (2021), So, what is a physics-informed neural network?.
• Maarten Grootendorst (2025), A visual guide to LLM agents.
• Kokotajlo et al. (2025), AI 2027.

Source code:
https://github.com/anthology-of-data-science/lecture-recent-developments-ai

Daniel Kapitan, Recent development in AI.
This work is licensed under CC BY-SA 4.0

Learning objectives

Learn

• Physics-informed machine learning and AI for scientific discovery
• Current issues in the ‘AGI arms race’
• LLM agents

Do

• Case study of possible application of PIML
• Exploration of future scenarios on societal and ethical impact of AI
• Hand-on session on exploring toolkit for LLM agents and MCP

Agenda

1. Lecture Physics-Informed Machine Learning

   • presentation
   • example
     
     

2. To AI Or Not To AI: exploring critical uncertainties for the coming years
   
   

3. LLM Agents (link)

To AI or not to AI?

Path dependence

source: Wikipedia

Path dependence is a concept in the social sciences, referring to processes where past events or decisions constrain later events or decisions. It can be used to refer to outcomes at a single point in time or to long-run equilibria of a process. Path dependence has been used to describe institutions, technical standards, patterns of economic or social development, organizational behavior, and more.

This means:

• The future development of AI is affected by the path it has traced out in the past
• past events or decisions affect future events or decisions in significant or disproportionate ways, through mechanisms such as increasing returns, positive feedback effects and other mechanisms.

Step 1: exploring critical uncertainties with

• Which uncertainties do we see in the development of AI in the coming 3 years?
  
  

• Which of these uncertainties are critical causing a major disruption in the labour market?

  • Scope: “white collar” desk-bound jobs such as call centre agents, programmers, accountants, copy-writers etc. which is roughly 40% of all jobs in the EU
  • Major disruption: more than 10% in jobs

Step 2: exploring scenarios with the two most critical uncertainties

Agenda

1. Lecture Physics-Informed Machine Learning

   • presentation
   • example
     
     

2. To AI Or Not To AI: exploring critical uncertainties for the coming years
   
   

3. LLM Agents (link)

Thanks for your attention.