Studies & Observations
we generate a steady stream of self-initiated and client-facing research. whenever possible, when it serves to educate and inform,we make it available on our website or publish it elsewhere.
ChatGPT 4o Deep Research and Midstream No-Notice Capability Suspension
A researcher discovered that ChatGPT's internal research tool had been disabled mid-project without notice or explanation. This led to a prompt-response chain on service reliability and trust-loss relating to arbitrary removal of paid-for capabilities.
Chasing Waterfalls: Cascade Effects and the AI Competencies Debate
Notes from the MIMOSA PROTEST experiment on LLM benchmarks, sandbox resources, preference cascades, and the professional pitfalls of chasing waterfalls.
What Happens When You Take LLMs Out of the Lab
We’ve been running an in-house experiment, MIMOSA PROTEST, on the premise that technology end-users, the ones outside the design labs, aren’t “lay users” or “non-technical” users. They’re technical specialists in their own fields of work….
Semi-Automated Media Monitoring on the Cheap
We’ve been experimenting with low-cost, makeshift approaches to automated media monitoring and open-source research. The result looks and feels more like Gibson’s unevenly distributed future than the Jetsons’ bright and shiny version- but it’s a reliable system.
Iterative Prompting and Concept Analysis for Brand Development
This case study covers an informal in-house brainstorming exercise using ChatGPT. It shows an iterative, 15-point prompt-and-response chain, in a non-technical format that can be easily replicated or adapted by non-technical users.
Clippings: Data Hoarders and the Data Rescue Project
One of the great - and ironic - benefits of a state as well documented as the US, is that efforts to remove, delete, or ignore its vast pools of data inevitably inspire widespread resistance and countermeasures…
LLMs, Search Engines, and Reasoning by Analogy in ChatGPT
This research note observes that AI work is fundamentally experimental, and discusses technology assisted research, user purpose and search object in AI prompting, and distinctions between large language models and search engines.
This research note is part of a series on AI benchmarking and cross-pollinated methods.
Applying COSTAR to Daily Reporting in ChatGPT
This research note is part of an occasional series on AI benchmarking and cross-pollinated methods. The note addresses COSTAR prompting, working with different prompt configurations, modifying COSTAR to prioritise information targets, and the pitfalls of putting the horse before the cart.
RCT and 5P Prompt Configurations
This research note is part of an occasional series on AI benchmarking and cross-pollinated methods. The note addresses different prompt configurations for role-context-task (RCT prompting) and prime-persona-privacy-product-polish (5P prompting) tasks.
Governance, Infrastructure, Information (DIIS Talking Points 5)
This research note on governance and infrastructure is the fifth in a series on evidence controls and digital technologies. The series develops points raised at the Danish Institute for International Studies, and research now part of a larger international collaboration on information and communications technology.
Forensic Expectations and the CSI Effect (DIIS Talking Points 4)
This research note on forensic expectations and aesthetics is the fourth in a series on evidence controls and digital technologies. The series develops points raised at the Danish Institute for International Studies, and research now part of a larger international collaborative project on information and communications technology.
The UNITAD Mission in Iraq (DIIS Talking Points 3)
This research note on the case of post-war Iraq and the UN Investigative Team for the Accountability of Daesh (UNITAD) is the third in a series on evidence controls and digital technologies. The series develops points raised at the Danish Institute for International Studies, and research now part of a larger international collaborative project on information and communications technology.
The OSINT Landscape in Ukraine (DIIS Talking Points 2)
This research note on the case of wartime Ukraine is the second in a series on evidence controls and digital technologies. The series develops points raised at the Danish Institute for International Studies, and research now part of a larger international collaborative project on information and communications technology.
Introduction, Themes, and Case Studies in Evidence Control (DIIS Talking Points 1)
This research note introduces and is the first in a series on evidence controls and digital technologies. The series develops points raised in a talk given at the Danish Institute for International Studies event, on the role of digital devices in politically contentious environments. The research is now part of a larger project on information and communications technology, in collaboration with colleagues in the United Kingdom, Sweden, Denmark, and the Netherlands.