Dr Steve J. BickleyResearcher, Engineer & Entrepreneur

Behavioural economics · AI · Complex systems

Current research themes

My research examines how people make decisions and how behaviour, technologies and institutions evolve together. I combine behavioural economics, complexity science and computational methods to connect individual decisions with organisational change and wider social and economic outcomes.

Human–AI interaction and evaluation

How well can generative AI represent human values, attitudes, preferences and decisions? I investigate synthetic populations, or silicon sampling, through comparisons with human evidence, sensitivity analysis and cross-model evaluation.

This includes questions of cultural alignment, population heterogeneity, public opinion and organisational judgement. I’m interested in which research questions synthetic respondents can help address, and how to recognise the variation and uncertainty that their responses may miss.

Selected work
SurveyLM technical report ↗
Comparing human and synthetic data in service research · Journal of Services Marketing ↗
Using reminders with reward opportunities to reduce spectator no-show behavior · Journal of Business Research ↗
AI prediction and human understanding · The Conversation ↗

Technology adoption, work and capability

Why do some workers and organisations adopt new technologies productively while others face barriers, risks or uneven benefits? I combine behavioural and institutional analysis with labour market evidence, skill networks and models of adaptation.

I led the national study with Future Skills Organisation (FSO) of generative AI adoption by Australian workers, engaging a working group spanning government, employers, unions and professional bodies. Current research extends these questions to occupational mobility, capability evolution, professional identity and the ways institutions shape technological change.

Selected work
FSO's national AI adoption study & employer toolkit →
AI and big data in sustainable entrepreneurship · Journal of Economic Surveys ↗
AI copilots, work & professional identity →

Finding evidence in archives and interviews

I build research software for searching and analysing documents, archives and interviews. It combines semantic retrieval and language models with structured extraction and knowledge graphs, keeping results linked to the original material.

Projects include historical Spanish Inquisition records, scientific interviews through Web of Stories, and AI-assisted qualitative analysis. Researchers need to be able to check what the software has extracted against the source.

Selected work
Archival research & scientometrics platforms →
RiskWise AI-assisted qualitative analysis workflow →

Complex systems and behavioural economics

I study technological evolution as a complex adaptive process: people learn and make decisions, organisations develop capabilities, and institutions shape which changes persist. These micro, meso and macro levels influence one another through networks, feedback and adaptation.

My PhD brought behavioural economics and complexity science together to study interacting agents and limited information. That perspective now informs my work on workforce transitions, AI, digital platforms, public health, financial markets and mobility.

Selected work
Bridging Complexity and Behavioural Economics · PhD thesis ↗
Artificial intelligence in the field of economics · Scientometrics ↗
Cognitive architectures for artificial intelligence ethics · AI & SOCIETY ↗

How I work

Methods & approaches

I choose methods around the research question, the scale of the system and the available evidence. My work often combines empirical analysis with computational modelling and applied software development, with particular attention to validation, robustness, practical implementation and the assumptions underlying the results.

  • Behaviour and evidence: survey design, experiments, behavioural economics, statistical analysis and econometrics.
  • Systems and complexity: agent-based modelling and simulation, network analysis, scientometrics, geospatial and mobility analysis.
  • AI and computational methods: machine learning, generative AI, synthetic populations (a.k.a. silicon sampling), computational experiments, model validation and benchmarking, structured information extraction and human-in-the-loop workflows.
  • Research and project delivery: requirements definition, stakeholder engagement and co-design, feasibility and technology evaluation, business case development, risk management, quality management systems and ISO 9001-informed project management and delivery.

Across these activities, I work with interdisciplinary teams and external partners to translate technical methods, research evidence and practical implications for audiences across industry, academia, policy and the public.

Software & tools

  • Research, analysis & modelling: Python, R, Stata, MATLAB, SQL, NetworkX, Gephi, QGIS and ArcGIS.
  • AI development & agentic systems: LLM application development, agentic and multi-agent workflows, retrieval-augmented generation (RAG), semantic search, knowledge graphs, tool/function calling, structured outputs and Model Context Protocol (MCP) integrations.
  • Application development: Django REST Framework, React, Streamlit and REST API development.
  • Infrastructure & deployment: PostgreSQL, Docker, Git/GitHub, GitHub Actions (CI/CD), Linux/Bash, high-performance computing (HPC) and AWS (including Lightsail).

The Constrained Methods Matrix

My PhD thesis, Bridging Complexity and Behavioural Economics (QUT, 2023), developed a framework for choosing methods for complex problems. The Constrained Methods Matrix relates nine broad families of analytical methods to the number of causes involved, the scale of their effects, and the complexity and uncertainty of the information available.

The framework grew out of work with Professor Benno Torgler, building on ideas from Marvin Minsky, Aaron Sloman and Push Singh. Its boundaries are deliberately soft: statistical, qualitative and case-based reasoning may be useful together, and collecting different data can make other approaches feasible.

I also consider feedback between levels of a system. Individual behaviour can change an organisation, while organisational rules shape what individuals can do; these effects unfold at different speeds. The relevant focal level might be a workplace, a sector or a region.

View the methods matrix and feedback diagram
Nine families of analytical methods arranged by number of causes, scale of effect, information complexity and information uncertainty. Methods include statistical analysis, neural networks, qualitative reasoning, fuzzy logic, analogy, symbolic logic and case-based reasoning.Browse Figures 1–4
Figure 1 · The Constrained Methods Matrix, extended version. Match analytical approaches to the system’s scale, causes and available information. Thesis p. 13, adapted from Bickley & Torgler (2023). Thesis and source context ↗
Upper, focal and lower levels of a system are connected by feedforward arrows and feedback loops operating over hours and days, days and weeks, and weeks, months and years.View larger
Figure 5 · Nested feedback loops. Thesis p. 56. The chapter draws on hierarchical systems work by Simon (1962) and Ceddia et al. (2013). Thesis and source context ↗

In the thesis, I applied this perspective to public health, financial markets, mobility and the development of AI. Read the complete thesis ↗

Research reach

Citations and readership

My work is read across economics, AI, public health and related fields. The figures below identify the source and the date it was last checked.

Google Scholar citations
738Checked
Google Scholar h-index
9Checked
Field-weighted citation impact
4.70SciVal · snapshot

Google Scholar is checked weekly; the dates record the last successful refresh. View my current Google Scholar profile ↗. SciVal is a dated snapshot, rather than a live counter.

Peer-reviewed publication venues

My peer-reviewed research has appeared in the following journals:

AI & Society (SJR 1.504, Q1); European Journal of Political Economy (SJR 1.131, Q1); Globalization and Health (SJR 1.692, Q1); Journal of Business Research (SJR 3.498, Q1); Journal of Economic Surveys (SJR 2.483, Q1); Journal of Services Marketing (SJR 1.536, Q1); Research in International Business and Finance (SJR 1.509, Q1); Safety Science (SJR 1.443, Q1); Scientific Reports (SJR 0.893, Q1); Scientometrics (SJR 1.077, Q1); Technology Analysis & Strategic Management (SJR 0.824, Q2); and Tourism Economics (SJR 0.998, Q1).

2025 SJR snapshot · Score and best subject quartile · Checked 29 October 2026

Journal-level SJR scores and quartiles, verified against public publisher and university research records. Quartiles vary by subject category; the best SJR quartile is shown here. This is a dated snapshot. Explore SCImago Journal & Country Rank ↗

Policy and public engagement

Where the research is being used

Employer resources and government evidence

The Future Skills Organisation study led to a national report, employer workshops and resources for organisations introducing generative AI. My work with IP Australia linked job advertisements, firm information and patent records to study AI investment.

FSO study and employer resources →
Download FSO’s SME AI Coach ↗
IP Australia: Drivers of AI Investment ↗

Citations in policy-facing research

My research has been cited in IZA discussion papers on AI and its economic implications. Two examples are:

These citations show my work contributing to wider policy discussions. The commissioned projects provide more direct examples of my research being developed into resources for government and employers.

News, media and public discussion

Selected reporting and commentary on my research, with further mentions and republications below.

Other outlets that have featured my work

The Quantum Insider; Communications Bureau Netherlands–Germany; Hacker News; MSN; Yahoo! News; Makronom; PsyPost; IT Boltwise; Desde Abajo; Le Point; The Conversation; El Mostrador; and E&N.

More coverage and public discussion

Writing for a wider audience

I’ve written about synthetic respondents for The Conversation, co-authored an opinion article on AI and software maintenance for The Mandarin, and written about COVID-19 and the environment for the Nature research community.

Read the articles and engagement figures

Working across disciplines

Research collaborations

I work with researchers, students and practitioners across universities, government and industry. The collaborations range from joint papers and student supervision to software development, commissioned studies and resources for employers.

Universities and research organisations

At QUT, I work with ARC BITA, the Energy Transition Centre, the Centre for Data Science and eResearch. Collaborators are also based at the University of Western Australia, Universidad de Oviedo, Universidade da Coruña, UNSW, Maastricht University, Universität Tübingen, the University of Kassel and Griffith University.

Other research connections include CREMA, the Laboratory for Innovation Science at Harvard, WHU – Otto Beisheim School of Management, the University of Fribourg and CSIRO / Data61. My international exchanges have also involved the University of Economics Ho Chi Minh City and RMIT Vietnam.

Government, industry and health

Selected collaborators and project partners include Future Skills Organisation, IP Australia, the Queensland Department of the Premier and Cabinet, Queensland Department of Transport and Main Roads, Inspired Education Australia, Verbrec, Suncorp, Stryker Trauma, Impect, Metro South Health and Princess Alexandra Hospital.

The FSO project also drew on a national working group with representatives from government, major employers, unions, industry and small business bodies, including the Digital Transformation Agency, Commonwealth Bank, Woolworths Group, the Australian Council of Trade Unions and COSBOA. Project, team and working group details →

Want to discuss a potential research collaboration with me?

Research figure

Expanded research figure

Open full-resolution image