AI & research
AI in environmental journalism
My FTC 2025 talk, with the recording, published paper and context for that stage of the research.
Talk · 2025 · EnglishWatch the talk & explore sources
Science · AI · Technology
TechJournal
Understanding technology before it becomes everyday life.
I’m Torsten. I want to understand what new ideas can do, where their limits lie and what they mean for everyday life. On TechJournal, come along with me — into the studio, to the computer and out into the field.
Current research
Preprint · not peer reviewed
v1 · submitted 15 September 2026AIJIM · Publication Authority
What still holds when an AI-assisted claim changes?
A person approves a text. Then a source or sentence changes. What does that approval still cover? Our public preprint explores how a claim, its evidence and a human decision can remain bound to the same publication state. This concerns accountability, not automatic truth.
A thought experiment
One sentence changes. What about approval?
State A · Claim
At this sampling point, today's reading was below the threshold.
- Evidence
- Measurement A · one place · one moment
- Human decision
- Approval for state A
The decision applies to this exact claim with this exact evidence.
Fictional illustration of the research question. Not a real review, a running AIJIM system or proof that the claim is true.
7:24 min · EnglishThe first research film · FTC 2025
From an observation to a research question.
AI, people and environmental journalism in Mallorca. My original talk, connected to the published paper and an account of its limits.
Film & research recordThe topics
Curious? Start here.
Science and technology are all around us. Explore my topics, from my own research to developments, ideas and people far beyond it.
01Current AI
What can it actually do?
New models, agents and tools: exploring their capabilities, mistakes and practical uses.
How I approach a claim02Trust without magic
What makes good evidence?
Plenty of things sound convincing. Sources, tests and clear examples help us take a closer look.
Explore the sources03UTS, research & AIJIM
How does a question become research?
Inside my PhD at the University of Technology Sydney: research papers, experiments and questions still waiting for answers.
Explore my research04AI infrastructure & studio
What goes into an experiment?
Local AI, computers, cameras, sound and voice systems: the equipment and work behind the scenes.
Look inside the studio05AI in journalism
How is AI changing journalism?
From the first source to publication: what helps, what remains uncertain and which decisions do people make?
Watch a research example06NamicGreen & real consequences
What is happening on the ground?
Environmental observations, people and fieldwork in Mallorca. NamicGreen remains an active project, with its stories and future developments part of TechJournal.
Watch existing reportsOn YouTube
Watch. Wonder. Think further.
The channel already has a history: research, the environment, people and places. The existing NamicGreen films remain part of the shared TechJournal channel.
Existing videos & collectionsEnvironment & fieldwork
Water & environment · Mallorca
Observations, environmental questions and reports from the island.
PlaylistOpen playlist
People & stories
Reports & background
NamicGreen’s history and the people and places behind it.
PlaylistOpen playlist
Nature & travel
Mallorca & Seychelles
Nature and landscapes from the existing film archive.
PlaylistOpen playlist
From the archive
An introduction to AIJIM
An early look at the idea behind AIJIM, reflecting the project at that time. The September preprint's record explains the current research.
Archive videoWatch the archive video
Taking a closer look
Good questions take us further.
Enthusiasm for technology and critical questions belong together. These four steps help make sense of a claim — and leave room to change our minds.
What is the claim?
Make it specific: what is supposed to work, for whom and under which conditions?
What supports it?
Look at sources, observations and tests. A compelling demo is a starting point; its conditions matter too.
What remains open?
Mistakes, missing data and counterexamples can help us understand.
What follows from it?
What possibilities emerge? What can we try — and where do we need to look more closely?
The story continues
Film → Record → Revisit
Records for the FTC talk, BIS paper and September preprint already bring together original sources and limits. Not every paper has a film yet. New findings should lead back to the original story, rather than silently overwrite it.
Watch
The film introduces a question and shows people, technology or an observation.
Explore
The accompanying record brings together sources, context, relevant research artifacts or cases, and open questions.
Revisit
New findings and corrections connect back to the original story.
My research
Research gets interesting long before it is finished.
In my PhD at UTS, I explore how we can trace and question AI-assisted claims: which sources support them, what has changed and what did a person decide? AIJIM, SHARK and NamicGreen play different roles.
My work spans an AI model for environmental journalism, an Evidence-First architecture for organisational decision support, and a machine-checkable specification of publication conditions. Two published conference chapters and a preprint that has not been peer reviewed document this research trajectory.
They examine different objects: findings from the early environmental study do not establish the performance of the later reference model or protocol. The three research records present each contribution, its authorship, sources and evidence boundaries separately.
Research approach
AIJIM Protocol
Making sources, claims and accountability traceable
Research & specification development
AIJIM explores how sources, changes and human decisions can remain connected to a published claim. Public papers and versioned research artifacts document its development.
Explore papers and sourcesExperiment
AIJIM SHARK
My experimental editor environment
Experimental
In SHARK, I explore ways of representing sources, changes and human decisions when writing with AI. Which approaches help, and where they add effort, remain open questions.
Explore SHARKEnvironment & fieldwork
NamicGreen
An independent environmental project
Active project
What remains of an environmental photograph when someone else takes over the case? NamicGreen aims to keep observations, evidence, decisions and the next inspection together. Originating in Mallorca, it is being developed for Mallorca and New South Wales, Australia. An old finding does not establish today's conditions.
Explore NamicGreenSources & publications
For anyone who wants to dig deeper.
Explore published papers, preprints and research artifacts from AIJIM. Each source has its own scope and status. Follow the links to read the originals.
Conference chapter
AIJIM: A Scalable Model for Real-Time AI in Environmental Journalism
- Online since 16 October 2025
- SourceSpringer · FTC 2025 · DOI chapter 26Read the originalOpen research record
- Context
- Early AIJIM research in environmental journalism. Its results concern the approach described there and do not establish the effectiveness of today’s SHARK.
Conference chapter
An Evidence-First Reference Model for Trustworthy AI in Business Information Systems
- Online since 21 May 2026
- SourceSpringer · BIS 2026 · DOI chapter 7Read the originalOpen research record
- Context
- A reference model for traceable work with AI. Its conceptual contribution is distinct from later practical validation.
AIJIM specification paper
Making AI-Assisted Claims Independently Challengeable: Publication Authority and a Protocol for Falsifiable Publication Records
- v1 · submitted 15 September 2026
- SourceTiltack et al. · arXiv:2609.17631v1Read the originalOpen research record
- Context
- How can AI-assisted claims stay connected to their evidence, human authorization and later corrections? This paper proposes and examines a protocol for that purpose. Its results apply to the models and tests studied; they establish neither a claim’s truth nor effectiveness in field use.
Software & research artifacts
AIJIM Publication Authority — Specification and Reproducibility Artifacts
- Versions 0.1.0 / 0.2.0
- SourceZenodo · one collection, two versions
- Context
- Versioned specification and reproducibility artifacts. Version 0.2.0 adds a research addendum on corrections. These are not two separate publications and do not establish journal acceptance or general effectiveness.
About me
I want to know what is behind it.
I’m Torsten Olivi Tiltack, a PhD Candidate at the University of Technology Sydney. I’m interested in scientific questions as much as trying things out: what can we trace, demonstrate and explain clearly?
TechJournal is my personal channel for that curiosity. My research is part of it, alongside science from elsewhere, new technologies, questions in journalism and the people working with them.
Being transparent
- My own research, observations and editorial interpretation are identified as such.
- AI assistance is disclosed where it materially shapes a contribution.
- UTS is my personal academic affiliation. TechJournal is an independent personal channel.
TechJournal
Bring your curiosity. Keep thinking.
Science, AI and new technology — clearly explained and explored in practice. Join me on YouTube.