Automatic synthesis of CAE
3 min talk and demo of a research tool to synthesis Assurance 2.0 cases. Given at Assure 2024.
AI, security, safety, resilience, trust.
This is site brings together short notes, unpublished work in progress, as well as signposting and summarising of other work I have been involved with.
I can be reached at reb@0point6fte.com or r.e.bloomfield@citystgeorges.ac.uk
There is a page with links to Assurance 2.0 papers, mainly by John Rushby and myself. There are also a number of posts about work in progress and opinions on topical issues.
Latest posts summarised below
What might go wrong with AI safety and security?
This post looks at what can be learned from engineering critical systems. It summarises a paper written from FAISC and on arxiv.
Resilience – Type 1 and Type 2 (coming soon)
Complex systems will fail, so they need to be resilient. Type 1 deals with known issues we engineer for and Type 2 those we exclude and the unknown unknowns.
Complete list of posts
- Automatic synthesis of CAE
- What might go wrong with AI safety and security?
- Resilience – Type 1 and Type 2 (coming soon)
- Assurance 2.0 News
If you have any paper, tool or book recommendations, please let us know. I’m particularly interested in how to get a handle on decision making around high impact AI enabled systems: issues of how confident we need to be and how we can model and evaluate systems with deep uncertainty.
What next? 