ArticlePDF Available

1(45) (2025): International Journal of Innovative Technologies in Social Science CITATION Nataliya Onishchenko, Oleksii Kostenko, Dmytro Zhuravlov. (2025) AI Technologies to The Question of The "Policy" of Legal Regulation at The Present Stage. Essential and Instrumental Factors

Authors:
  • Independent Researcher

Abstract

The article examines the essential characteristics of rulemaking activity in the context of modern challenges and priorities, which is analyzed with due regard for the instrumental and essential typologizing elements. It is noted that one of the priority areas for the development of rulemaking at the present stage is to consider the "achievements" of the latest technologies related to artificial intelligence. It is emphasized that rulemaking at all levels should ensure human rights and freedoms (in particular, this refers to the improvement of veteran policy at the present stage, its forms, and methods).
International Journal of
Innovative Technologies in
Social Science
e-ISSN: 2544-9435
Scholarly Publisher
RS Global Sp. z O.O.
ISNI: 0000 0004 8495 2390
Dolna 17, Warsaw,
Poland 00-773
+48 226 0 227 03
editorial_office@rsglobal.pl
ARTICLE TITLE
AI TECHNOLOGIES TO THE QUESTION OF THE "POLICY" OF
LEGAL REGULATION AT THE PRESENT STAGE. ESSENTIAL AND
INSTRUMENTAL FACTORS
ARTICLE INFO
Nataliya Onishchenko, Oleksii Kostenko, Dmytro Zhuravlov. (2025) AI
Technologies to The Question of The "Policy" of Legal Regulation at The Present
Stage. Essential and Instrumental Factors. International Journal of Innovative
Technologies in Social Science. 1(45). doi: 10.31435/ijitss.1(45).2025.3200
DOI
https://doi.org/10.31435/ijitss.1(45).2025.3200
RECEIVED
29 November 2024
ACCEPTED
19 January 2025
PUBLISHED
27 January 2025
LICENSE
The article is licensed under a Creative Commons Attribution 4.0
International License.
© The author(s) 2025.
This article is published as open access under the Creative Commons Attribution 4.0 International License (CC
BY 4.0), allowing the author to retain copyright. The CC BY 4.0 License permits the content to be copied, adapted,
displayed, distributed, republished, or reused for any purpose, including adaptation and commercial use, as long
as proper attribution is provided.
1(45) (2025): International Journal of Innovative Technologies in Social Science
e-ISSN: 2544-9435
1
Introduction.
One of the interesting points that we would like to emphasize is, of course, the conceptual approaches,
fundamental principles, and communication practices that have been and are used in relation to rulemaking
activities. They have been thoroughly studied in domestic and international legal thought from different angles
and in different temporal coordinates, and the capabilities and potentials of artificial intelligence deserve
additional study in this context, and this article is designed to address them. The purpose of the article is to
emphasize the need to focus on new emphases and new aspects of rulemaking, its existing and possible
typological classification characteristics.
At certain stages of legal development, this issue was dealt with by legal theorists, including S.V.
Bobrovnyk, T.O. Didych, M.I. Kozyubra, N.M. Onishchenko, O.V. Petryshyn and representatives of branch
directions: N. Kuznetsova, O. Kot, Z. Zahinei-Zabolotenko, and others. Nevertheless, many questions remain,
1(45) (2025): International Journal of Innovative Technologies in Social Science
e-ISSN: 2544-9435
2
particularly in the area of artificial intelligence. Today, such issues cannot be considered non-negotiable and
require consensus-building among the scientific community.
1. Problems and unexplored elements of typology.
According to the universal dictionary of the Ukrainian language, typology is a classification of objects or
phenomena according to common features [1, с. 750]. A large explanatory dictionary also defines typology as a
type of scientific systematization, classification of something according to common features [2, с. 602]. The
typology of states is a classification of groups of states according to common features, which determine the
characteristics of these groups of states. The typology of legal (legal) systems is a classification of legal systems
according to their social characteristics, which determine their socio-essential characteristics [3, с. 912].
After encyclopedic research, we want to emphasize at least those classification accents that have not
been researched or insufficiently researched, or which, unfortunately, have been left out of the attention of
representatives of theoretical thought.
In our opinion, such typological elements can be instrumental or essential features (associated with the
corresponding meanings dictated by modern conditions). By the way, it is this approach, taking into account
instrumental practices and essential characteristics, that could be useful when considering the problems of
adaptation of Ukrainian legislation to the legislation of the European Union.
2. Instrumental mechanisms of typology.
For example, in Belgium, the Institute of Social Law is developing a rulemaking system called SOLON
to determine the quality of legislation, which includes a substantive criterion (content of legal acts) and a
formal criterion (structure, formatting, etc.) at the request of the Government. Moreover, such an idea is not a
novelty, as similar systems are already in use in the Kingdom of the Netherlands (LEPA, OBW) and in the
Italian Republic (Lexidit, Lexeditior IRI_AI, Arianna, Norma). There is even a whole science called
“legimatics” that studies and explores the possibilities of computer technology in the field of rulemaking.
Scientists of the Institute of Social Law highlight the following advantages of the SOLON system:
1) it helps lawmakers avoid mistakes;
2) helps to create draft laws faster and in a more productive way
3) applies guidelines to rulemaking that prevent lawmakers from changing the sequence of actions;
4) plays an important role, as sectoral experts often do not have a legal background.
In turn, rulemaking in the Kingdom of the Netherlands is notable for the existence and use of algorithms
for creating legal rules that transform formulas into legal rules, which greatly simplifies and speeds up the
rulemaking process (rulemaking), but requires special mathematical and computer science knowledge. This
experience has been successfully implemented in domestic higher education institutions, particularly, the
experience of KNUPE, where, before the full-scale invasion, the information (technical) factor of creating and
testing legal norms was actively developed.
In this context, special attention can be paid to rulemaking in Canada, where such activity is considered
an “art” rather than a routine activity that employs about 100 people.
In our opinion, these rules should include the need to involve both representatives of legal doctrine and
legal practice. Certain attempts to do so have been made repeatedly in various government institutions,
including the Ministry of Justice of Ukraine. However, today there is an urgent need to create a single center
for planning and studying rulemaking activities, which was repeatedly discussed even before the full-scale
invasion of Ukraine by Russia. And what makes this need (for rule-making activities) even more urgent today.
3. Essential features of typology.
It is somewhat more difficult to study issues that are related to the essential substantive aspect of
normative design. The philosophical significance of the essential characteristics of normative design activity
was discussed, in particular, in the studies of R.K. Bergeron, where several rules of normative design activity
were distinguished:
1) fixation of the key philosophical position on the intercorrespondence of content to the forms of the project;
2) the author of the rules of rule-making proposes to take into account the time factor of such activities,
believing that insufficient time can affect the quality of development and the very quality of the regulatory legal act;
3) E.K. Bergeron draws attention to the importance of a professional approach to such activities (the
factor of professionalization);
1(45) (2025): International Journal of Innovative Technologies in Social Science
e-ISSN: 2544-9435
3
4) attention must necessarily be focused on the possible negative result of normative drafting (the factor
of negativity of rulemaking);
5) it is also stated that the rules of normative drafting are not and will never be absolute (the factor of
deabsolutization), since they cannot cover the entire set of development and design of normative texts;
6) teleological dominants of normative design (teleological factor) must be considered [4, с. 24].
In addition, it should be noted that the study of the experienced practices of different legal systems does
not exclude those tasks that become quite prominent concerning the domestic community. In particular, the
attention of foreign donors is focused on the following stages of regulatory drafting in Ukraine:
1) Creation of a Ukrainian style of rule-making, which will be able to prevent existing inconsistencies
in legislation caused by the work of different groups of lawmakers and various sources of law, in particular,
we are talking about alternative normative acts, the number of which is growing rapidly;
2) Normalization of current consultations with foreign advisers, whose rule-making technique is based
on the rules of rule-making inherent in the jurisdiction of other countries;
3) Assisting the Ukrainian Government in bringing Ukrainian legislation in line with the legislation of
the European Union;
4) Deepening the existing understanding in public opinion of the principles of democracy,
constitutionalism, ethics, and fair governance and applying them in the law-making process;
5) Fight corruption by preventing contradictions and eliminating gaps in the current legislation [5].
4. Ukraine's unique experience in normative design.
The above considerations can also serve as an interesting basis for distinguishing among the essential
characteristics of rulemaking activities the experience of Ukraine, which is currently not available in any other
country in the world. Such provisions should be supported by illustrative conclusions. In particular, this can
be emphasized in view of the work currently being done in the field of veterans' policy. This refers to the draft
Law of Ukraine “On the Basic Principles of State Veteran Policy”. It is this draft that is being worked on by a
working group established under the Ministry of Veterans Affairs of Ukraine. We would like to focus on the
following provisions. Public policy, as we understand it, consists of many facets: social, legal, etc. One of
these hypotheses, and one of the primary ones today, is veterans' policy. Veterans' policy is usually considered
in the context of veterans' integration into civilian life. However, we understand today that after the Victory, a
two-pronged process will take place: society will integrate veterans, and veterans will adapt society to life in
new conditions and realities, given the proximity of the state sponsoring terrorism. In addition, the level of
“non-legal consciousness” and “non-legal culture” existing in Russia deserves special attention in this context.
It is impossible to influence these processes in one day, one week, or even one year. Therefore, it is systematic
work, which will consist of appropriate educational practices, starting with schools, higher education
institutions, enterprises, organizations, etc., that can be organized by veterans, their representatives, and
veterans' centers [6].
Another aspect that we would like to emphasize today, demonstrating the need for substantive
etymologization when adapting Ukrainian legislation to EU legislation, should be the principles on which
veteran policy in Ukraine is implemented.
1. The rule of law: ensuring the priority of the rights and freedoms of veterans, members of their families,
and family members of deceased veterans.
2. The principle of humanism: respect for the dignity of veterans.
3. The principle of democracy: the possibility for veterans, their family members, and family members
of deceased veterans to participate in the formation and implementation of the state veterans' policy.
4. Barrier-free: ensuring unimpeded access of veterans, members of their families, and family members
of deceased veterans to various spheres of human activity.
Thus, after illustrating the substantive characteristics required by rulemaking, we would like to
emphasize once again that legal acts should correspond to the realities of life, and rulemaking practices should
take into account the current context of the global legal order. This means that rulemaking at all levels, from
supreme public authorities (general rulemaking), central executive authorities (departmental rulemaking), local
executive authorities (local rulemaking) to rulemaking by heads of enterprises, institutions, and organizations
(local rulemaking), should effectively ensure human rights, freedoms, and legitimate interests.
It is clear that even this range of problems (many of which remain beyond the scope of this article)
requires not only new approaches but also the solution of the tasks set. We will allow ourselves to focus on the
so-called “intellectually artificial” use of certain modern technologies.
1(45) (2025): International Journal of Innovative Technologies in Social Science
e-ISSN: 2544-9435
4
5. Innovative approaches to scientific research and normative design practice.
At the same time, the realization of the above tasks is radically changing its vector in the direction of
modern immersive technologies, Metaverse, artificial intelligence (AI), blockchain, cryptography, robotics,
and quantum technologies. Science and Technology Revolution 5.0 has not only stimulated technological
breakthroughs but also launched the development of new social relations, the transformation of morality and
law, ethics and philosophy.
On November 30, 2022, OpenAI released ChatGPT, which provided scientists with a wide range of
modern tools for generating the latest scientific research, paradigms, theories, models, and discoveries. Under
such conditions, rulemaking can acquire radically different methods and algorithms, one of which is considered
promising, namely, the method of modelling the socio-legal evolution of civilization and law.
For example, R.M. Montgomeri's research analyses possible trajectories of human civilization
development under the influence of AI. The research is based on stochastic differential equations (SPDEs) to
model probable scenarios in time-space. The model takes into account the impact of AI and visualizes a wide
range of potentially balanced prospects for the future of civilization with significant probabilities in both
positive and negative directions [7]. In addition, a team of Indian scientists proposes a new metaheuristic
algorithm for social evolution by studying the processes of optimizing human social learning (SELO). The
research is based on the newest class of optimization algorithms - socially inspired algorithms that calculate
the social propensity of people to adapt to the manners and behaviour of others through observation and
learning [8,9]. However, these studies generally do not consider the alternative retrospective simultaneous
development of society and law.
6. Methodology for Predicting the Future Evolution of Humanity and Law Using Dynamic AI Models.
Our proposed method of modelling the socio-legal evolution of civilization and law is to develop a dynamic
AI model of human development and law using AI and other immersive technologies, big open data based on
historically reliable sources and legal documents of all eras and social systems. This model is an innovative,
alternative construction independent of personal influence, in which two parallel “family trees” of social relations
and law are simultaneously built historically, and their recursive “historical gene” of fundamental definitions is
formed from a simple basic legal structure to its modern correlation in space and time.
The proposed dynamic AI model of human development and law and the method of modelling the socio-
legal evolution of civilization are aimed at creating an innovative dynamic model that takes into account the
historical, social and legal aspects of human development through the use of AI, immersive technologies and
big data (fig.1).
Fig.1. Dynamic AI model of human development and law
The algorithm of the dynamic AI model of human development is as follows:
1. Analysing data and historical sources, legal documents of different eras (laws, codes, court decisions,
international agreements); studying social constructions (traditions, rituals, behavioural norms, and social
structures); rethinking historical facts confirmed by archaeological and documentary sources. AI algorithms
1(45) (2025): International Journal of Innovative Technologies in Social Science
e-ISSN: 2544-9435
5
provide automatic processing of this data, analysing to identify patterns, trends, and key points of
transformation of the social and legal system.
2. Creation of “family trees” of society and law, namely parallel “family trees” of social relations,
including all stages of society's evolution from primitive to modern globalized systems, family structures,
social hierarchies, and legal systems, from basic legal structures such as customary law to complex legal
systems of the present. These “trees” are not static - they are built dynamically with the help of AI algorithms
that identify key “nodes” (periods of significant change) and reflect the relationship between social processes
and legal norms.
3. Formation of a conditional “historical gene” of law based on data analysis, which contains: the basic
legal structure of primary legal norms that determined social relations at a certain stage of history; evolutionary
modifications of the “historical gene” of law and changes that occurred under the influence of economic,
political, technological or cultural factors; modern correlation and current versions of legal norms that are the
result of a long evolution. This approach allows us to visualize both elements of law and social norms and their
transformation over time, preserving their main features or adapting to new conditions.
4. Immersive technologies for interactive analysis, such as virtual reality (VR) or augmented reality
(AR), are integrated into the model to create an interactive environment to immerse users in virtual
reconstructions of legal systems from different eras, simulate alternative scenarios of law evolution based on
variable parameters (e.g., what would have happened if a certain historical event had not occurred), and study
the cause-and-effect relationship between social change and legal reforms.
A dynamic AI model of human development and the socio-legal evolution of civilization and law should
perform the following main tasks
- distinguishing legal constructions from primary legal norms of national law, jurisdictions of other
states, and international legislation from the definition-artifact to modern legal constructions
- re-engineering of national law to dilute it from legal normative constructions that are duplicated, have
different logical and legal content, create funnels of “legal uncertainty” or scenarios of legal contradictions;
- to help researchers and lawmakers to formulate legal acts, norms and definitions in synergy with the
Constitution of Ukraine, fundamental rights, freedoms and ethical principles of humanity;
- functioning of systemic AI as both a “law auditor” and a “supervisory module”, with the function of a
scientific and advisory structure that studies social transformation and develops proposals for prompt legal
response to changes in social relations in Ukraine and cross-border space, forecasting legal trends, assessing
the impact of modern decisions on the future evolution of law, preserving historical heritage, and popularizing
knowledge about human development.
The main feature of the dynamic AI model of human development is its independence from personal
influence. The algorithms analyze data objectively, without bias, creating scientifically grounded
reconstructions. At the same time, the model allows considering the multiplicity of factors that influenced the
evolution of law and society, providing a deep and multifaceted analysis. Such a system has the potential to
become a universal tool for studying historical dynamics, understanding the present, and predicting the future.
7. Technologies for creating algorithmic copies and simulacra: impact on social and legal evolution.
Today, in our opinion, there are three fundamental studies that can have a huge impact on the development
of society and law. These are technologies for creating algorithmic copies of any social groups developed with the
use of AI technologies, which give grounds to assert the real possibility of implementing a model of civilizational
and legal modelling of the positive development of Ukraine and civilization as a whole. LLM GPT language models
can be used as a proxy for human cognition at the aggregate level and as universal windows into human thinking -
technologies can create an algorithmic copy of a person and an algorithmic copy of any social group. Now, for
sociological research, it is no longer relevant to interview physical respondents; this procedure has been transferred
to algorithms that imitate given social groups or individuals [10].
The first study. In 2023, scientists developed the “Chinese Room of Increased Complexity” technology
to create algorithmic copies of citizens of any country [11]. This was followed by the Wuhan experiment to
predict the US presidential election in 2024 based on the analysis of the AI model of preferences of simulacra
rather than people. The simulation was conducted 90 days before the elections and predicted the victory of the
new president with a 99% probability, a difference of 3 units for each party [12].
The second study. In 2024, researchers conducted the Stanford Simulacrum Experiment, in which they
created a thousand simulacra of the individual consciousness of “typical” Americans. That is, real Americans
were selected to represent the US population in terms of age, gender, education, and political views. The main
1(45) (2025): International Journal of Innovative Technologies in Social Science
e-ISSN: 2544-9435
6
tool of the joint research of Stanford University and Google DeepMind, as in the case of the Wuhan experiment,
is generative AI of large speech models (ChatGPT-4o). According to the results, the simulacra predicted the
reaction of their real human prototypes to the GSS test with 85% accuracy, and the simulacra gave results that
almost did not differ from the reactions of their human prototypes (correlation coefficient of 0.98) [13].
The third study. A universal computational model of human cognition, Centaur, was created by retraining
the open language model Lama 3.1 70B on a new large-scale dataset called Psych-101. Psych-101 is capable of
accurately predicting and modelling any human behaviour in any experiment from any field that can be described
in natural language, i.e. AI models become indistinguishable from humans in their behaviour in any situation and
circumstance related to research, planning, and training. In other words, AI models can not only operate with our
languages without us noticing but also behave like intelligent entities similar to us [14].
8. Simulacra and AI in Modeling the Interaction of Society, State, Social Groups, and Citizens in
the Future of Ukraine as a Subject of the World Legal Space.
The use of AI in legal systems is an important stage in the development of modern society that can
significantly affect various aspects of the judicial process, including access to justice, decision-making
efficiency, and the accuracy of legal document analysis. A high level of automation can lead to fair results not
only at the initial stages of the legal decision-making process but also throughout the entire procedural vertical,
reducing the destructive nature of the “human factor”. The introduction of AI can lead to the so-called technical
and legal positivism, when decisions will be made solely based on algorithms that take into account only
formal legal norms and facts, leaving aside the human factor, moral values, and individual approach to each
case. This raises serious concerns about the preservation of the principle of justice and humanity in justice, as
automation can lead to the dehumanization of the judicial process.
We proposed two main “family trees” to create the model. In fact, the dynamic AI model of human
development and the socio-legal evolution of civilization and law is multifunctional, and by analogy, it is
possible to build a socio-legal “family tree” to model the field of veteran policy and legislation, using the best
legal artifacts of the past and modern practices. Artificial intelligence is able to model the psychophysical
archetypes of veterans, make forecasts for this social group, create models of interaction between society, the
state, and veterans, and, based on the analysis of their simulations, form an appropriate regulatory framework.
These changes in society and law in the field of veterans' policy will be adapted to reality as much as possible
based on the results of the simulation, which will greatly reduce the time and resources spent on rulemaking
and improve the quality of laws. Simulators and AI can become a positive tool that will ensure high-quality
interaction between people and the state.
Artificial intelligence has the potential to significantly improve the efficiency and accessibility of law,
provided that it is implemented carefully and prudently.
The digital transformation of law is quite positive and pragmatic, especially given the inevitability of
integration and cross-border processes of Ukraine's entry into the global legal space as an entity that, after the
war, may become a “typical model” for restructuring the state system, social, legal and economic conditions
of the functioning of a modern state in the emerging digital world. In view of this, the proposed Method of
modeling the socio-legal evolution of civilization and law and the dynamic AI model of humanity and law
development will allow Ukraine to conduct a large-scale independent and objective audit of law, significantly
optimize the active part of the state's legal field, start the process of creating legal and moral norms of the
present, relevant and realistic definitions, and, most importantly, provide a reliable legal shield for the
Constitution of Ukraine, democracy, fundamental human rights and freedoms.
Conclusions.
For a long time, the thesis has been discussed that legislation needs constant updating, and the process
of drafting regulations should finally undergo positive changes, which means its effectiveness and
effectiveness. The thesis after the full-scale invasion of the Russian Federation into Ukraine seems especially
prominent. In turn, this means a call to the scientific community to distinguish between the processes that
existed "before" and "after", to apply new approaches, using the experience that existed, the new experience
that exists "from the present" in Ukraine. In particular, such an approach to typology can be considered the
allocation of instrumental (model), technical characteristics in normative design. We will talk about the forms
of normative legal acts and content characteristics, which also require constant updating and cannot be covered
by dogmatic (once and for all given) approaches to normative design practices. At the same time, we especially
emphasize, emphasize, emphasize that this does not mean separating the "form" from the "content" of the legal
1(45) (2025): International Journal of Innovative Technologies in Social Science
e-ISSN: 2544-9435
7
act, but on the contrary, contributes to the improvement of normative drafting by studying both the best
practices of the world level, and the elaboration and provision of perfect normative forms of a new experience
of legal development in Ukraine.
REFERENCES
1. Universalnyi slovnyk ukrainskoi movy. Ternopil: Navchalna knyha Bohdan, 2005. 848s.
2. Velykyi tlumachnyi slovnyk. Suchasna ukrainska mova vid A do Ya. Donetsk: TOV VKF «BAO», 2008. 704 s.
3. Velykyi entsyklopedychnyi yurydychnyi slovnyk / Za red. Yu.S.Shemshuchenka. 2-he vyd., pererobl. i dopovn.
Kyiv: «Iurydychna dumka», 2012. 1020 s.
4. Chornolutskyi R. Pravyla normoproektvannia R.K. Berzherona ta yikh znachennia dlia stanovlennia y rozvytku
natsionalnoi systemy normoproektuvannia v Ukraini. National law journal: Teory and Practice. 2015. C.24.
5. Normoproektuvalna diialnist: vymohy teorii ta praktyky. Pravo ta yustytsiia. 2018. № 4 (5). S. 15-22.
6. Onishchenko N.M. Pravova politika yak skladova yevrointegraciinoyi politiki Ukrayini: vikliki, zagrozi, perspektivi.
(Za materialami dopovidi na zasidanni Prezidiyi NAN Ukrayini 15 travnya 2024 r.) Visnik Nacionalnoyi akademiyi
nauk Ukrayini. 2024. № 7. S. 57-61
7. Montgomeri R.M. (2024). Stokhastychnе modelyuvannya maybutn'oho lyuds'koho suspil'stva za vplyvu
konkuruyuchykh АІ. Zhurnal matematychnykh tekhnolohiy ta obchyslyuval'noyi matematyky. URL: https ://doi.org
/10.33140/jmtcm.03.07.07.
8. Kumar M., Kulkarni A., Satapathy S. (2018). Socio evolution & learning optimization algorithm: A socio-inspired
optimization methodology. Future Gener. Comput. Syst., 81, 252-272. URL:
https://doi.org/10.1016/j.future.2017.10.052.
9. Perera Y.S., Ratnaweera D.A.A.C., Dasanayaka C.H., Abeykoon C. (2023). The role of artificial intelligence-driven
soft sensors in advanced sustainable process industries: A critical review. Engineering Applications of Artificial
Intelligence, 121. URL: https://doi.org/10.1016/j.engappai.2023.105988.
10. Argyle L.P., Busby E.C., Fulda N., Gubler J.R., Rytting C., Wingate D. (2023). Out of One, Many: Using Language
Models to Simulate Human Samples. Political Analysis, 31(3), 337-351. DOI: 10.1017/pan.2023.2.
11. Durmus E., Nguyen K., Liao T.I., Schiefer N., Askell A., Bakhtin A., Chen C., Hatfield-Dodds Z., Hernandez D.,
Joseph N., Lovitt L., McCandlish S., Sikder O., Tamkin A., Thamkul J., Kaplan J., Clark J., Ganguli D. (2023).
Towards Measuring the Representation of Subjective Global Opinions in Language Models. arXiv:2306.16388. DOI:
10.48550/arXiv.2306.16388.
12. Jiang S., Wei L., Zhang C. (2024). Donald Trumps in the Virtual Polls: Simulating and Predicting Public Opinions
in Surveys Using Large Language Models. arXiv:2411.01582. DOI: 10.48550/arXiv.2411.01582.
13. Park J.S., Zou C.Q., Shaw A., Hill B.M., Cai C., Morris M.R., Willer R., Liang P., Bernstein M.S. (2024). Generative
Agent Simulations of 1,000 People. arXiv:2411.10109. DOI: 10.48550/arXiv.2411.10109.
14. Binz M., Akata E., Bethge M., Brändle F., Callaway F., Coda-Forno J., Schulz E. (2024, October 27). Centaur: a
foundation model of human cognition. URL: https://doi.org/10.31234/osf.io/d6jeb.
ResearchGate has not been able to resolve any citations for this publication.
Preprint
Full-text available
Large language models (LLMs) may not equitably represent diverse global perspectives on societal issues. In this paper, we develop a quantitative framework to evaluate whose opinions model-generated responses are more similar to. We first build a dataset, GlobalOpinionQA, comprised of questions and answers from cross-national surveys designed to capture diverse opinions on global issues across different countries. Next, we define a metric that quantifies the similarity between LLM-generated survey responses and human responses, conditioned on country. With our framework, we run three experiments on an LLM trained to be helpful, honest, and harmless with Constitutional AI. By default, LLM responses tend to be more similar to the opinions of certain populations, such as those from the USA, and some European and South American countries, highlighting the potential for biases. When we prompt the model to consider a particular country's perspective, responses shift to be more similar to the opinions of the prompted populations, but can reflect harmful cultural stereotypes. When we translate GlobalOpinionQA questions to a target language, the model's responses do not necessarily become the most similar to the opinions of speakers of those languages. We release our dataset for others to use and build on. Our data is at https://huggingface.co/datasets/Anthropic/llm_global_opinions. We also provide an interactive visualization at https://llmglobalvalues.anthropic.com.
Article
Full-text available
With the predicted depletion of natural resources and alarming environmental issues, sustainable development has become a popular as well as a much-needed concept in modern process industries. Hence, manufacturers are quite keen on adopting novel process monitoring techniques to enhance product quality and process efficiency while minimizing possible adverse environmental impacts. Hardware sensors are employed in process industries to aid process monitoring and control, but they are associated with many limitations such as disturbances to the process flow, measurement delays, frequent need for maintenance, and high capital costs. As a result, soft sensors have become an attractive alternative for predicting quality-related parameters that are 'hard-to-measure' using hardware sensors. Due to their promising features over hardware counterparts, they have been employed across different process industries. This article attempts to explore the state-of-the-art artificial intelligence (Al)-driven soft sensors designed for process industries and their role in achieving the goal of sustainable development. First, a general introduction is given to soft sensors, their applications in different process industries, and their significance in achieving sustainable development goals. AI-based soft sensing algorithms are then introduced. Next, a discussion on how AI-driven soft sensors contribute toward different sustainable manufacturing strategies of process industries is provided. This is followed by a critical review of the most recent state-of-the-art AI-based soft sensors reported in the literature. Here, the use of powerful AI-based algorithms for addressing the limitations of traditional algorithms, that restrict the soft sensor performance is discussed. Finally, the challenges and limitations associated with the current soft sensor design, application, and maintenance aspects are discussed with possible future directions for designing more intelligent and smart soft sensing technologies to cater the future industrial needs.
Article
Full-text available
We propose and explore the possibility that language models can be studied as effective proxies for specific human subpopulations in social science research. Practical and research applications of artificial intelligence tools have sometimes been limited by problematic biases (such as racism or sexism), which are often treated as uniform properties of the models. We show that the “algorithmic bias” within one such tool—the GPT-3 language model—is instead both fine-grained and demographically correlated, meaning that proper conditioning will cause it to accurately emulate response distributions from a wide variety of human subgroups. We term this property algorithmic fidelity and explore its extent in GPT-3. We create “silicon samples” by conditioning the model on thousands of sociodemographic backstories from real human participants in multiple large surveys conducted in the United States. We then compare the silicon and human samples to demonstrate that the information contained in GPT-3 goes far beyond surface similarity. It is nuanced, multifaceted, and reflects the complex interplay between ideas, attitudes, and sociocultural context that characterize human attitudes. We suggest that language models with sufficient algorithmic fidelity thus constitute a novel and powerful tool to advance understanding of humans and society across a variety of disciplines.
Preprint
The promise of human behavioral simulation--general-purpose computational agents that replicate human behavior across domains--could enable broad applications in policymaking and social science. We present a novel agent architecture that simulates the attitudes and behaviors of 1,052 real individuals--applying large language models to qualitative interviews about their lives, then measuring how well these agents replicate the attitudes and behaviors of the individuals that they represent. The generative agents replicate participants' responses on the General Social Survey 85% as accurately as participants replicate their own answers two weeks later, and perform comparably in predicting personality traits and outcomes in experimental replications. Our architecture reduces accuracy biases across racial and ideological groups compared to agents given demographic descriptions. This work provides a foundation for new tools that can help investigate individual and collective behavior.
Preprint
In recent years, large language models (LLMs) have attracted attention due to their ability to generate human-like text. As surveys and opinion polls remain key tools for gauging public attitudes, there is increasing interest in assessing whether LLMs can accurately replicate human responses. This study examines the potential of LLMs, specifically ChatGPT-4o, to replicate human responses in large-scale surveys and to predict election outcomes based on demographic data. Employing data from the World Values Survey (WVS) and the American National Election Studies (ANES), we assess the LLM's performance in two key tasks: simulating human responses and forecasting U.S. election results. In simulations, the LLM was tasked with generating synthetic responses for various socio-cultural and trust-related questions, demonstrating notable alignment with human response patterns across U.S.-China samples, though with some limitations on value-sensitive topics. In prediction tasks, the LLM was used to simulate voting behavior in past U.S. elections and predict the 2024 election outcome. Our findings show that the LLM replicates cultural differences effectively, exhibits in-sample predictive validity, and provides plausible out-of-sample forecasts, suggesting potential as a cost-effective supplement for survey-based research.
Preprint
Establishing a unified theory of cognition has been a major goal of psychology. While there have been previous attempts to instantiate such theories by building computational models, we currently do not have one model that captures the human mind in its entirety. Here we introduce Centaur, a computational model that can predict and simulate human behavior in any experiment expressible in natural language. We derived Centaur by finetuning a state-of-the-art language model on a novel, large-scale data set called Psych-101. Psych-101 reaches an unprecedented scale, covering trial-by-trial data from over 60,000 participants performing over 10,000,000 choices in 160 experiments. Centaur not only captures the behavior of held-out participants better than existing cognitive models, but also generalizes to new cover stories, structural task modifications, and entirely new domains. Furthermore, we find that the model’s internal representations become more aligned with human neural activity after finetuning. Taken together, Centaur is the first real candidate for a unified model of human cognition. We anticipate that it will have a disruptive impact on the cognitive sciences, challenging the existing paradigm for developing computational models.
Article
The paper proposes a novel metaheuristic Socio Evolution & Learning Optimization Algorithm (SELO) inspired by the social learning behaviour of humans organized as families in a societal setup. This population based stochastic methodology can be categorized under the very recent and upcoming class of optimization algorithms-the socio-inspired algorithms. It is the social tendency of humans to adapt to mannerisms and behaviours of other individuals through observation. SELO mimics the socio-evolution and learning of parents and children constituting a family. Individuals organized as family groups (parents and children) interact with one another and other distinct families to attain some individual goals. In the process, these family individuals learn from one another as well as from individuals from other families in the society. This helps them to evolve, improve their intelligence and collectively achieve shared goals. The proposed optimization algorithm models this de-centralized learning which may result in the overall improvement of each individual's behaviour and associated goals and ultimately the entire societal system. SELO shows good performance on finding the global optimum solution for the unconstrained optimization problems. The problem solving success of SELO is evaluated using 50 well-known boundary-constrained benchmark test problems. The paper compares the results of SELO with few other population based evolutionary algorithms which are popular across scientific and real-world applications. SELO's performance is also compared to another very recent socio-inspired methodology-the Ideology algorithm. Results indicate that SELO demonstrates comparable performance to other comparison algorithms. This gives ground to the authors to further establish the effectiveness of this metaheuristic by solving purposeful and real world problems.
Suchasna ukrainska mova vid A do Ya. Donetsk: TOV VKF «BAO
  • Slovnyk Velykyi Tlumachnyi
Velykyi tlumachnyi slovnyk. Suchasna ukrainska mova vid A do Ya. Donetsk: TOV VKF «BAO», 2008. 704 s.
Berzherona ta yikh znachennia dlia stanovlennia y rozvytku natsionalnoi systemy normoproektuvannia v Ukraini. National law journal: Teory and Practice
  • R Chornolutskyi
  • R K Pravyla
Chornolutskyi R. Pravyla normoproektvannia R.K. Berzherona ta yikh znachennia dlia stanovlennia y rozvytku natsionalnoi systemy normoproektuvannia v Ukraini. National law journal: Teory and Practice. 2015. C.24.
Pravova politika yak skladova yevrointegraciinoyi politiki Ukrayini: vikliki, zagrozi, perspektivi. (Za materialami dopovidi na zasidanni Prezidiyi NAN Ukrayini 15 travnya 2024 r.) Visnik Nacionalnoyi akademiyi nauk Ukrayini
  • N M Onishchenko
Onishchenko N.M. Pravova politika yak skladova yevrointegraciinoyi politiki Ukrayini: vikliki, zagrozi, perspektivi. (Za materialami dopovidi na zasidanni Prezidiyi NAN Ukrayini 15 travnya 2024 r.) Visnik Nacionalnoyi akademiyi nauk Ukrayini. 2024. № 7. S. 57-61