The possibilities and limitations of synthetic research. Choose a topic and explore the findings and sources.
16 articles
A population-scale simulated-user evaluation infrastructure that tests AI systems and digital products with 8.3 billion persona records and 1,010 reusable tasks.
In a 400-trial controlled behavioral study, the assigned behavior was expressed or correctly suppressed in 91.5% of trials.
How accurately LLM agents imitate human opinions when given demographic information and belief networks.
Using demographic information alone did not produce meaningful human–LLM alignment.
How LLMs behave in social interactions: the role of group identity, reciprocity and chain-of-thought reasoning.
LLMs do not act solely out of self-interest; they also lean towards social welfare and reciprocity.
Rational decision-making by LLMs in financial scenarios under uncertainty: improving decision accuracy and transparency through factor profiles and historical analogy.
The DEFINE framework achieved higher accuracy and F1 scores than the alternative methods (accuracy 29.6%, F1 23.7%).
The capacity of LLMs to imitate human agents in behavioural economics experiments: the effect of endowing models with different attributes and the reproducibility of results.
GPT-3 qualitatively reproduced the findings of behavioural economics experiments run with human participants.
Using LLMs as subpopulation representative models (SRMs): potential for measuring public opinion and a risk–benefit assessment.
LLMs can serve as powerful tools for capturing public opinion and representing subpopulations.
The role of LLMs in agent-based modelling and simulation (ABMS): transforming the simulation paradigm through human-like intelligence and behaviour.
Planning, memory and feedback are among the core mechanisms that LLM-based agents use in simulations.
Learning and imitating individual opinions with an LLM-based Doppelgänger model: accuracy, heterogeneity, and single-device feasibility.
The Doppelgänger model replicated individual opinions with high accuracy.
The capacity of LLM agents to imitate human behaviour across social systems, and the modelling of complex social dynamics with Generative Agent-Based Models (GABMs).
LLMs can generate human-like behaviours such as fairness, cooperation, and adherence to social norms.
LLMs’ ability to perceive and interpret user personas: the impact of cultural context and the capacity to reconstruct demographic profiles.
GPT-3.5 and GPT-4 showed strong alignment across the three India-centred personas analysed.
Calibration of LLMs across demographic groups: human alignment and transferability using Human Mimicry Calibration (HMC).
Human Mimicry Calibration (HMC) significantly improved alignment between LLM responses and human data.
Persona usage in LLMs: the distinction between Role-Playing and Personalization, application domains, and personality assessment methods.
The fundamental distinction in persona research is between Role-Playing (the LLM adopting a persona) and Personalization (the system adapting to the user’s persona).
The capacity of LLMs to simulate human behaviour across classic experiments and diverse demographic profiles: model scale, demographic variation, and hyper-accuracy effects.
LLMs successfully replicated several known patterns of human behaviour.
Evaluating how biased LLMs are towards brands and how consistent their answers are.
All LLMs showed consistent (transitive) brand preferences, producing the ordering Apple > Samsung > Huawei.
The ability of LLMs to imitate human preferences and the effects of language and chain-of-thought methods.
LLMs (GPT-3.5 and GPT-4) were more impatient than humans; GPT-4’s discount rates were far above human levels.
How LLM-based search affects efficiency, accuracy and user perception compared with traditional search, and the role of confidence cues.
LLM-based search cut task completion time in half (1.6 min vs. 3.4 min).
No articles match this search. Try another term or select “All”.
