Laypeople’s Understanding of Artificial Intelligence

Background and Problem Statement. Research on artificial intelligence (AI) demonstrates considerable breadth in the application of this term. According to sociological data, approximately half of Russians report confident knowledge about artificial intelligence. However, no generally accepted understanding of AI has yet emerged; the term is a collective one. Philosophical and psychological accounts of AI are built upon its comparison with human intelligence, which has a rational nature but is inseparable from non-rational aspects of the mind. The understanding of AI as such by its users has been insufficiently studied from a psychological perspective. Objective. This study aims to construct a psychological characterization of laypeople’s understanding of artificial intelligence. Methods. An online survey of 34 participants (79.4% women, 20.6% men) aged 15 to 40 years (M = 22). Statistical processing was performed in the R statistical software package. For each question, a proportion test (prop.test) with a chi-squared criterion (df = 1) was used to assess the achieved significance level of the null hypothesis that the relative frequency equals 0.5, and a 95% confidence interval (0.95 CI) was calculated for the relative frequency of fixed response options. Results. A positive attitude toward AI was found in the studied sample, based on personal experience of its use. Participants noted the risks of using AI while considering them manageable by the user, compared seeking AI assistance with other information sources, and affirmed that users might be interested in discussing even love-related questions with AI. In responses to an open-ended question, participants described various motives for seeking AI assistance: difficult situations where people cannot solve problems independently; convenience and practicality in solving various tasks; play motivation (curiosity); neutrality and psychological safety of AI consultation; and possible difficulties or complete lack of human communication. Conclusions. The obtained results align with sociological research data and allow for a psychological characterization of users’ understanding of AI as a functional, instrumental, and interactive phenomenon applicable to all spheres of life. Whether functionality, instrumentality, and interactivity of AI are synonyms, three independent characteristics, or are sublated by a fourth requires further research.

Introduction

Artificial intelligence (AI) is a phenomenon whose understanding by users has been studied far from sufficiently to date. The diversity of what may be classified as AI is such that producing any systematic literature review is extremely difficult. D. V. Ushakov rightly points out that AI is a collective term (Ushakov, 2023). Does the fact that energy companies use AI to optimize load distribution allow us to classify as AI users everyone who turns on a light? From a psychological perspective, the answer to this question is not as straightforward as it might initially appear.

Sociological research shows that “the concept of ‘artificial intelligence technologies’ has become firmly embedded in everyday life. Over the past two years, the proportion of Russians declaring confident knowledge about such technologies (have heard of them and can explain them) has grown from 36% to 50%. Another 44% possess superficial knowledge; they have heard of AI technologies but cannot yet explain what they are (2022: 51%). Over the years, it is the depth of knowledge that has grown, while its breadth appears to be approaching its limit” (VCIOM, 2024). V. Yu. Menovshchikov offers the following definition of AI: “in the broadest sense, it is intelligence demonstrated by machines, particularly computer systems” (Menovshchikov, 2025, p. 14), noting that “many artificial intelligence applications are no longer perceived as artificial intelligence” (Menovshchikov, 2025, p. 15).

In substance, Menovshchikov’s understanding aligns well with the philosophical one: “Various research programs of ‘artificial intelligence’ have emerged: 1) the creation of computers capable of performing functions traditionally attributed to human intellectual activity; 2) attempts to model human intelligence itself based on modeling the neural substrate (neurocomputers); 3) the creation of artificial self-learning devices capable of evolution” (Suvorov, 2010).

Philosophical and scientific psychological understandings of human intelligence emphasize its rational nature and relate it to non-rational aspects of the psyche: “rational thinking subject to the laws of logic; it opposes the non-rational spheres of the psyche: emotions, imagination, will, etc.” (Suvorov, 2010). In psychology, it is generally accepted that “the nature and direction of intelligence cannot be understood without taking into account personal qualities, needs, emotional states, etc. On the other hand, personal maturity largely depends on the degree of a person’s understanding of what is happening, which is linked to the level of intellectual development” (Kholodnaya, 2002).

The philosophical understanding of rationality demonstrates, first, the universality of this term: “Rationality (from Latin ratio, reason) is a term symbolizing one of the key themes of philosophy, a fundamental problem whose solution is determined by the general content of a particular philosophical-methodological concept. The problem consists in clarifying the meaning of ‘reasonableness’ as predication (of being, action, relation, goal, etc.). <…> The theme of rationality problematizes virtually all major spheres encompassed by modern philosophical thought” (Porus, 2010). Second, the process of rationalization closely resembles the development of AI: “the sequential overcoming of the elements of nature, culture, and the human soul (psyche) and their replacement by logically ordered systems of practices following the principle of efficiency <…> should be understood precisely as a process requiring permanent creative efforts, an unceasing balancing of heterogeneous principles; it is unattainable in any ‘automatic’ mode of following a model or system of rules” (Panarin, 2010).

Thus, the rational nature of human intelligence, which creates, uses, and understands AI, is inseparable from less rational aspects of the psyche. And we observe the same in relation to AI.

AI research offers alluring prospects: human-computer interaction, brain-computer interfaces, breakthroughs in multimodal and multidimensional data research and data extraction, and the simulation of human emotions (Zhao et al., 2022). It is not even necessary for AI to actually possess emotions, experience, or consciousness. Language that meets social expectations is sufficient to embed AI in joint action based on a relational interpretation of interaction with AI. Then AI begins to help people reflect, practice communication, or rehearse emotionally difficult conversations without replacing the human (Boyd & Markowitz, 2026). But the question of replacing the human turns out to be not so simple.

Responding to diagnostic instruments, AI successfully “outperforms” 50% of PhD holders and 90% of bachelor’s degree holders in an Arabic-language sample on measures of social intelligence (Sufyan et al., 2024). LLMs successfully discuss treatment issues with patients (Kaplan & Akyol, 2026) and “pass” medical exams better than humans (Chen et al., 2025). The diagnostic accuracy of LLM reasoning approximately corresponds to that of psychiatric residents (Jin et al., 2026). Variants of such discussion of treatment issues are proposed in which LLMs would improve in medicine by maintaining dialogue with each other (Sun et al., 2026) and learning through reinforcement (Gabriel, 2020). AI systems are indeed applied in medical decision-making, for example, decisions about medical evacuation (Hiratsuka et al., 2026).

It seems established that at the level of the nervous system, people behave differently toward humans than toward AI, but sometimes AI engages the brain even more than other humans (Harris, 2024).

The development of AI raises ethical questions. A distinction is made, for example, between explicit and implicit moral machines, depending on how directly moral values are accounted for in algorithms. It is desirable that the decisions of moral machines be compatible with human goals and priorities, but for this, one must know precisely what human goals and priorities are (Bonnefon et al., 2024). In addition, another difficulty arises. Most of the source code of AI models and agents is proprietary and classified. In other words, we often do not know how AI works. We have access only to input and output data, and we must infer the decision-making process. Many hidden functions are unavailable for study because we are not allowed to look at them, or they are not interpretable because we cannot understand them, or both (Taylor & Taylor, 2021). This problem can be addressed through experimental methods of psychology, as in the case of humans. Such is the approach of explainable artificial intelligence (XAI). It is no coincidence that D. V. Ushakov proposes the concept of “digital angels,” since “to ensure effective work as an assistant and coordinator, electronic systems must possess the maximum possible information, and only their unconditional reliability in preventing the dissemination of information and their focus on protecting the interests of a specific person can serve as a basis for them to perform their functions” (Ushakov, 2024, p. 88).

The question arises of how to algorithmically prevent a digital angel from transforming into a digital demon. Judging by the found definitions of AI, this is not easy.

On the SberPro website, without attribution, the following definition is given: “Artificial intelligence is the ability of computer systems to perform tasks requiring human intelligence: pattern recognition, decision-making, data analysis. Modern artificial intelligence is based on machine learning methods and the analysis of large data sets. The system does not act strictly according to predetermined rules but learns from examples, identifies recurring connections, and applies the acquired knowledge to solve new tasks. This fundamentally distinguishes AI from classical programming, where the system’s behavior is fully determined by the developer” (SberPro, 2026).

1. Kretov offers the following definition: “‘Artificial intelligence is a technical system capable of solving assigned tasks by adapting to external factors.’ That is, can any program that can solve tasks of a certain type be called artificial intelligence? Almost. AI must have the ability to learn” (Kretov, 2026).
2. Patrusheva and A. Pavlova define AI substantially differently, confirming Ushakov’s characterization: “it is not a tool or a program but a separate branch of computer science. AI specialists develop systems that analyze information and solve tasks similarly to how humans do” (Patrusheva & Pavlova, 2025).

Let us summarize the literature review.

First, AI has proven to be a phenomenon difficult to define. Authors more often designate the nature of the tasks it solves and the possibilities of its application than characterize it substantively.

Second, in scientific texts, AI is considered as a functional analogy of human intelligence, compared with it, and contrasted with it. As in the case of human intelligence, the rational nature of AI is ambiguous and difficult to understand. The mechanisms of human-created AI are often non-obvious to people.

Third, in the context of AI, scientific questions arise related to going beyond the boundaries of rationality. Like human intelligence, AI proves to be not fully separable from extra-intellectual phenomena: directionality, morality, the content of tasks being solved, limitations of received feedback, and others.

Choosing a context for empirical research on laypeople’s understanding of AI is not easy; the spectrum of possible options is too broad. The context of the present study was suggested by one event. On the eve of Valentine’s Day, February 13, 2026, the author received an email: “Your secret advisor in matters of love. Love is a delicate matter. Sometimes you need someone who will listen, help you find the right words, or simply cheer you up before an important step. In such moments, turn to Alisa AI, she will tell you how to find the path to your loved one’s heart. Alisa, help me out!” Given the holiday factor and the possible drawing of users’ attention to the advertised version of AI through other channels, it was decided to immediately conduct a study in the format of a short survey.

Method

An online survey with invitations to participate distributed through the VKontakte social network. The questions and participants’ responses are presented in full below.

Statistical processing was performed in the free statistical software package R (R Core Team, 2026). For each response option, a proportion test (prop.test) with a chi-squared criterion (X-squared) for one degree of freedom (df = 1) was used to assess the achieved significance level of the null hypothesis that the relative frequency equals 0.5, and a 95% confidence interval (0.95 CI) was calculated for the relative frequency.

Thirty-four individuals: 27 (79.4%) women and 7 (20.6%) men, aged 15 to 40 years (M = 22 years, SD = 5 years).

Participants reported the following regarding their education:

• Currently enrolled at a university: 22 (64.7%);
• Completed university: 6 (17.6%);
• Currently enrolled at school/college/technical school: 4 (11.8%);
• Completed school/college/technical school: 2 (5.9%).

Results

An unexpected result for the author was the dynamics of user response to the invitation to participate: all 34 responses were received within a single day, February 14, 2026, while starting from February 15, 2026, users stopped responding to invitations and personal messages requesting participation in the study.

Below, we present the questions and participants’ responses in full.

Question 1. Do you think an invitation to discuss love-related questions with artificial intelligence could interest any network users?

Thirty participants (88.2%) responded affirmatively, while four (11.8%) responded negatively. Statistical analysis showed that the relative frequency of affirmative responses significantly differs from 0.5 (X-squared = 18, p = .00002, 0.95 CI: 0.72-0.96), as does that of negative responses (X-squared = 18, p = .00002, 0.95 CI: 0.04-0.28).

Forty-four percent of participants responded to the open-ended question about the reasons for their opinion. Their responses can be tentatively divided into several groups. Some participants wrote that people turn to AI for help in difficult situations when they cannot solve problems on their own: “Some people cannot cope with problems with their partner on their own.” Others noted the convenience and practicality of AI in solving practical or behavioral tasks: “One can ask AI for advice, a date plan, places to go, think through literally everything with it… AI will conduct analytics and explain everything, it is a useful tool.” Some participants described what is essentially play motivation for using AI: “just for fun,” “interesting to learn AI’s opinion.” A separate group of responses indicated the neutrality and psychological safety of discussing love with AI: “With it, one can discuss any question, and it will never judge or laugh…”; “AI does not have a biased attitude and allows you to look at the situation from the outside.” Finally, some participants pointed to possible difficulties or even a complete lack of human communication: “There are very many lonely… people who simply have no one to talk to…” Responses emphasizing the irreplaceability of live communication were also voiced: “Communication with a living person is irreplaceable.”

Question 2. Do you think discussing love-related questions with artificial intelligence could be useful for a person?

The majority of participants (21 individuals, 60%) chose the option “Artificial intelligence is no worse than reading popular articles (posts, reels, etc.).” This result does not reach the level of statistical significance when testing the hypothesis of equality to 0.5 (X-squared = 1, p = .22, 0.95 CI: 0.44-0.77). The option “Artificial intelligence is more interesting than popular articles; one can interact with it” was chosen by 5 individuals (14.3%) (X-squared = 16, p = .00008, 0.95 CI: 0.06-0.32). The option “It is foolish to discuss anything with artificial intelligence” was chosen by 4 individuals (11.4%) (X-squared = 18, p = .00002, 0.95 CI: 0.04-0.28). Another 5 participants (14.3%) formulated their own answer (X-squared = 16, p = .00008, 0.95 CI: 0.06-0.32). Among the original formulations: “I do not consider it foolish, but I would not do it myself; one should primarily rely on one’s own feelings or on books by well-known psychologists. AI is not a psychologist and certainly cannot understand human feelings”; “The topic of love is too deep and individual to discuss with AI”; “Artificial intelligence can provide interesting, useful information, but it cannot experience all feelings, so it cannot replace articles written by humans, etc.”; “It can be useful, but as with everything (posts, reels, etc.), one needs to know moderation and understand that AI is not a reliable source of information”; “AI does not invent answers itself but constructs them based on the analysis of articles, scientific works, etc. It can produce a brief summary from various sources and draw conclusions, which is very useful, you do not have to read a ton of literature and delve into its essence yourself. The only problem is that Alisa AI is, to put it mildly, a mediocre neural network, so I do not know how useful it can be. I would not trust it.”

Question 3. Who do you think would be more interested in discussing love-related questions with artificial intelligence?

The response “Women” was chosen by 16 individuals (47.1%). This result does not statistically differ from 0.5 (X-squared = 0.03, p = .86, 0.95 CI: 0.30-0.65). The response “Both genders” was chosen by 12 individuals (35.3%) (X-squared = 2, p = .12, 0.95 CI: 0.20-0.54). The response “Men” was chosen by 5 individuals (14.7%) (X-squared = 16, p = .00008, 0.95 CI: 0.06-0.32). The response “Actually, no one will take it seriously” was chosen by 1 individual (2.9%) (X-squared = 28, p = .0000001, 0.95 CI: 0.002-0.17).

Question 4. Do you think artificial intelligence can actually help a person with love-related questions?

Opinions were divided exactly evenly: 17 individuals (50%) answered “Yes,” and 17 individuals (50%) answered “No.” In both cases, X-squared = 0, p = 1, 0.95 CI: 0.34-0.66.

Question 5. Whom can artificial intelligence help more?

The response “Both” was chosen by 17 individuals (50%) (X-squared = 0, p = 1, 0.95 CI: 0.34-0.66). The response “Men” was chosen by 6 individuals (17.6%) (X-squared = 13, p = .0003, 0.95 CI: 0.07-0.35). The response “Women” was chosen by 5 individuals (14.7%) (X-squared = 16, p = .00008, 0.95 CI: 0.06-0.32). The response “No one” was chosen by 6 individuals (17.6%) (X-squared = 13, p = .0003, 0.95 CI: 0.07-0.35). No one chose the option “I will formulate my own answer.”

Question 6. In your opinion, can artificial intelligence be used to manipulate or otherwise harm a person in love?

Twenty-four individuals (70.6%) responded affirmatively (X-squared = 9, p = .004, 0.95 CI: 0.58-0.89). Eight individuals (23.5%) responded negatively (X-squared = 9, df = 1, p = .004, 0.95 CI: 0.11-0.42). Another 2 individuals (5.9%) formulated their own answer; these responses were statistically counted as affirmative. Among them: “If a person misunderstood the advice, then yes, it can”; “It all depends on the person, their intentions, ability to think rationally, and level of trust in AI. If the question means harming a person who discusses love with AI, then indeed AI can sometimes give such advice, especially Alisa, that a person who accepts its responses without reflection as direct instructions and calls to action will ultimately harm themselves.”

Question 7. Do you yourself use artificial intelligence? (for any purpose: work, study, out of interest, etc.)

The response “Yes” was chosen by 28 individuals (82.4%) (X-squared = 13, p = .0003, 0.95 CI: 0.65-0.93). The response “Rarely” was chosen by 4 individuals (11.8%) (X-squared = 18, p = .00002, 0.95 CI: 0.04-0.28). The response “No” was chosen by 2 individuals (5.9%) (X-squared = 24.735, df = 1, p = .0000007, 0.95 CI: 0.01-0.21).

Question 8. How would you rate your experience of using artificial intelligence?

Thirty individuals (93.8%) rated their experience as generally positive (X-squared = 23, p = .000002, 0.95 CI: 0.77-0.99). Two individuals (6.3%) rated it as useless (does not solve my tasks) (X-squared = 23, p = .000002, 0.95 CI: 0.01-0.22). No one rated it as generally negative.

Discussion

Participants gave different responses, sometimes categorically preferring one option, sometimes choosing different options with equal probability. This can be considered an indirect indicator of response validity. The dynamics of response collection, all responses were received on a single “Valentine’s Day,” demonstrates both the significance of participants’ intrinsic motivation to participate in the study and the socio-psychological factor of the holiday fashion, in connection with which the study was conducted.

Judging by the confidence intervals of the relative frequencies of response options, the results are sufficiently generalizable, which corresponds to the characteristics of the sample.

In substantive terms, what draws attention is the categorically positive attitude of the studied sample toward AI, based on personal experience of its use. This characterizes only the studied sample; the general population, judging by sociological surveys (VCIOM, 2024; VCIOM, 2025; VCIOM, 2026), is less homogeneous in this regard. Our participants confirm VCIOM data that “young people under 34 have already quite actively and regularly incorporated neural networks into their lives” (VCIOM, 2025).

Consistent with VCIOM data that risks of the digital environment are perceived as probable but not fatal (VCIOM, 2026) is the participants’ clear preference for the response that AI can harm a person, with the caveat of one participant: “it all depends on the person, their intentions, ability to think rationally, and level of trust in AI.” That is, AI risk is largely controllable by users, even when turning to AI on such a matter as love.

The possibility of successful, from the users’ perspective, management of AI risks is also shown by responses to the question of whom AI can help more. In responses to the question of the functionality or usefulness of AI (“Can artificial intelligence actually help a person with love-related questions?”), participants’ opinions were evenly divided between “yes” and “no.” In combination with responses to the risk question, this shows the non-categorical nature of negative assessments of possible AI assistance in love-related questions. If half of the participants had been categorically convinced of AI’s uselessness, the distribution of responses to the questions about risks and about whom AI can help more would have differed from that obtained in the study.

When answering the question of who would be more interested in discussing love-related questions with AI, participants (predominantly female) most likely had themselves in mind, preferring the responses “women” and “both genders,” which confirms the accuracy of the responses.

On the question of whether discussing love-related questions with AI can be useful for a person, participants clearly preferred the response that AI is no worse than reading popular articles (posts, reels, etc.). Psychologically, such a response corresponds to the above-described responses to questions about the risks of turning to AI on love-related questions.

For the substantive characterization of laypeople’s understanding of AI, responses to the first questions are especially important. The rather categorical preference of participants for an affirmative answer to the question of whether users might be interested in discussing love-related questions with AI aligns well with the fact that 44% of participants formulated their own answer to the optional question of why they think so. These responses can be tentatively divided into several complementary groups. Some participants write that people turn to AI for help in difficult situations when they cannot solve problems on their own (“Some people cannot cope with problems with their partner on their own”). Others note the convenience and practicality of AI in solving practical or behavioral tasks (“One can ask AI for advice, a date plan, places to go, think through literally everything with it… AI will conduct analytics and explain everything, it is a useful tool”). Still others describe what is essentially play motivation for using AI (“just for fun”), which partly resembles the previous group but with a more pronounced emotional component. Some participants indicate the neutrality and psychological safety of discussing love with AI (“With it, one can discuss any question, and it will never judge or laugh…”; “AI does not have a biased attitude and allows you to look at the situation from the outside”). Finally, some participants speak of possible difficulties or even a complete lack of human communication (“There are very many lonely… people who simply have no one to talk to…”). To one degree or another, the identified groups of responses may characterize a single person in a real situation of turning to AI; they do not exclude each other. The study did not provide a sufficiently definitive answer to the question of what the possible motives for turning to AI on love-related questions are. But such an answer was not our goal. Discussing love-related questions with AI was more of an external occasion for the research invitation, connected with “Valentine’s Day.”

The subject of the study is laypeople’s understanding of AI. If the responses are interpreted as a whole, what comes to the fore is such a characteristic of AI understanding as its functionality, instrumentality, and interactivity. Precise terminological designation requires further conceptual and empirical elaboration. Figuratively speaking, users’ responses suggest that “however you ask, however you relate, however you understand, that is how AI will help.” This characteristic corresponds not to perceived usefulness of the technology (Gado et al., 2022) but points to an understanding of AI precisely as a technology. The results of the conducted study align with VCIOM data that “the role of neural networks is not limited to an instrumental function: already today, 14% of neural network users have turned to them for communication and emotional support” (VCIOM, 2025). Nevertheless, AI in users’ understanding remains precisely a technology, a tool. Perhaps this will change in the future, but today the situation is as described.

The second important result of our study is the absence of psychologically insurmountable boundaries for users regarding the spheres of AI application. The possibility of discussing with AI the most diverse questions, even such as love. How participants understand love is the subject of another, independent study. What matters is that love, too, can be discussed with AI, judging by the obtained results.

The obtained data allow us to embed laypeople’s understanding of AI into a broader theoretical framework. First, the understanding of AI as a functional and interactive phenomenon aligns with the Computers Are Social Actors (CASA) paradigm, according to which people unconsciously apply social heuristics to computers and AI systems because they call to mind similar social attributes as humans (Nass et al., 1994; Nass & Moon, 2000). Participants in our study demonstrate precisely such a “social” attitude toward AI: they discuss love-related questions with it and expect support and non-judgmental attitudes from it, which corresponds to key tenets of the CASA paradigm.

Furthermore, the Human-AI Interaction Model based on the Theory of Interactive Media Effects (HAII-TIME) (Sundar, 2020) proposes distinguishing between a “cue route” and an “action route” in the formation of trust in AI. Anthropomorphic cues activate users’ heuristic evaluations, which in turn influence trust and technology acceptance (Sundar, 2020). In our study, participants describing AI as a “tool” simultaneously endow it with social qualities (neutrality, absence of judgment, ability to “listen”), which reflects the operation of the “cue route,” the heuristic path of evaluation. At the same time, the preference for the response that “AI is no worse than reading popular articles” indicates that users do not overestimate the epistemic status of AI, maintaining critical distance.

An important theoretical basis for interpreting the obtained results is the three-factor theory of anthropomorphism (Epley et al., 2007), according to which people tend to anthropomorphize non-human agents to satisfy the need for social connection, to increase the effectiveness of interaction with an unfamiliar agent, and to explain its behavior. In our study, participants who indicate loneliness as a motive for turning to AI, as well as curiosity (“just for fun,” “interesting to learn AI’s opinion”), demonstrate both key mechanisms of anthropomorphization: social motivation and epistemic motivation. This aligns with data showing that anthropomorphic features of AI activate social presence, interpersonal warmth, and mind perception, making the ideal and social self more salient (Epley et al., 2007; Waytz et al., 2010).

However, anthropomorphization also carries risks. Research shows that users may form emotional connections with human-like AI, creating the risk of infringing on privacy and autonomy through over-reliance (Akbulut et al., 2024). In addition, anthropomorphic features can lead to “misplaced capabilities”: users begin to trust AI in matters where its competence is limited (Maeda, 2024). Our participants, however, demonstrate awareness of these risks: 70.6% believe that AI can harm a person in love-related matters, but the risk is perceived as manageable (“it all depends on the person…”). This aligns with the model of “calibrated trust,” according to which users with higher levels of AI literacy tend to evaluate system transparency and more accurately correlate trust with AI’s actual capabilities.

Finally, the understanding of AI as a tool applicable to all spheres of life aligns with the concept of AI literacy. Research shows that AI literacy is positively associated with trust in AI, with perceived transparency partially mediating this relationship. Our participants, being predominantly university students (64.7%), demonstrate a fairly high level of AI literacy: they not only use AI but also reflect on its limitations (“AI is not a reliable source of information”), which corresponds to contemporary understandings of critical AI use.

Conclusions

The conducted study is preliminary and pilot in nature. We deliberately conducted the survey with fairly “transparent” formulations of questions and response options in order to foreground the aspects of AI understanding that are conscious to users.

Our main result is the understanding by users of AI as a functional, instrumental, and interactive phenomenon applicable to all spheres of life.

At present, we are not prepared to answer the questions of whether the functionality, instrumentality, and interactivity of AI are synonyms, three independent characteristics, or are sublated (in G. W. F. Hegel’s terminology) by some fourth.

In the literature on AI, we have not encountered direct indications of such an understanding of AI by users, although the described aspects are implied in a number of articles. We note that in many scientific publications on AI, it is not even explicitly stated that the user cannot interact with AI directly but gains access to it only through a user interface.The psychological aspects of AI are quite naturally conceptualized “one step behind” its creation and development. There are many implied but not explicitly discussed aspects in works on this problematic.

Simple dichotomies are inapplicable to AI in users’ understanding: rational-irrational, intellectual-emotional, safe-risky, etc. The rationality and safety of AI in the user’s understanding depends on the user no less than on the device and principles of AI operation.  And it is quite possible that this psychological characterization of AI understanding will change as AI develops.

Let us return to the question posed at the beginning of the article: does the fact that energy companies use AI to optimize load distribution allow us to classify as AI users everyone who turns on a light? From a psychological perspective, the answer may be: in most cases no, but in some cases yes, depending on the person’s awareness of this action.

Acknowledgments: The author expresses sincere gratitude to M. M. Koroleva for discussing the idea and results of the study.

Ethics Statement: This study was conducted in accordance with the ethical principles of the Russian Psychological Society, including informed consent, confidentiality, and voluntary participation.

Author Responsibility: The author confirms that he is solely responsible for all aspects of the work, including its conception, execution, and interpretation, has verified the entire manuscript, and takes full responsibility for its integrity and accuracy.

CRediT Author Statement: V. K. Solondaev: Conceptualization, Methodology, Investigation, Statistical Analysis, Writing – Original Draft, Review & Editing.

Competing Interests: The author declares that the research was conducted in the absence of any commercial or financial relationships, as well as personal relationships, that could be construed as a potential conflict of interest or could have influenced the work reported in this manuscript.

Funding: This research received no external funding.

AI Use Statement: The author confirms that artificial intelligence tools were not used for generating scientific content, analyzing data, or writing the manuscript. All intellectual and analytical tasks were performed by the author independently.

References

References

1. Akbulut, C., Weidinger, L., Manzini, A., Gabriel, I., & Rieser, V. (2024). All too human? Mapping and mitigating the risks from anthropomorphic AI. Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, 7(1), 13-26. https://doi.org/10.1609/aies.v7i1.31613
2. Bonnefon, J.-F., Rahwan, I., & Shariff, A. (2024). The moral psychology of artificial intelligence. Annual Review of Psychology, 75, 653-675. https://doi.org/10.1146/annurev-psych-030123-113559
3. Boyd, R. L., & Markowitz, D. M. (2026). Artificial intelligence and the psychology of human connection. Perspectives on Psychological Science, 21(2), 192-220. https://doi.org/10.1177/17456916251404394
4. Chen, Y., Wen, B., & Zulkernine, F. (2025). A multiagent summarization and auto-evaluation framework for medical text: Development and evaluation study. JMIR AI, 4, e75932. https://doi.org/10.2196/75932
5. Epley, N., Waytz, A., & Cacioppo, J. T. (2007). On seeing human: A three-factor theory of anthropomorphism. Psychological Review, 114(4), 864-886. https://doi.org/10.1037/0033-295X.114.4.864
6. Gabriel, I. (2020). Artificial intelligence, values, and alignment. Minds and Machines, 30(3), 411-437. https://doi.org/10.1007/s11023-020-09539-2
7. Gado, S., Kempen, R., Lingelbach, K., & Bipp, T. (2022). Artificial intelligence in psychology: How can we enable psychology students to accept and use artificial intelligence? Psychology Learning & Teaching, 21(1), 37-56. https://doi.org/10.1177/14757257211037149
8. Harris, L. T. (2024). The neuroscience of human and artificial intelligence presence. Annual Review of Psychology, 75, 433-466. https://doi.org/10.1146/annurev-psych-013123-123421
9. Hiratsuka, V., Pritchard, A., Day, G., Thomas, T., Wark, K., Shane, A., Onders, R., Hernandez-Boussard, T., & Rice, B. (2026). Artificial intelligence/machine learning in Alaska Native healthcare systems: Symposium perspectives. International Journal of Circumpolar Health, 85(1), 2677426. https://doi.org/10.1080/22423982.2026.2677426
10. Jin, K. W., Rostam-Abadi, Y., Chaudhary, P., Garrett, M. A., Huang, A. S., Montelongo, M., Nagpal, C., Shei, J., Weathers, J., Zhang, J. S., Chen, Q., Kim, J., Malgaroli, M., Mathis, W. S., Rodriguez, C. I., Selek, S., Sharma, M. S., Pittenger, C., Yip, S. W., … Xu, H. (2026). Evaluating diagnostic accuracy and clinical reasoning of multiple large language models in psychiatry. medRxiv. https://doi.org/10.64898/2026.02.03.26345402
11. Kaplan, T. T., & Akyol, G. E. (2026). The role of large language models in dental diagnosis, decision-making, and communication: A systematic review. The Japanese Dental Science Review, 62, 57-67. https://doi.org/10.1016/j.jdsr.2026.01.002
12. Kholodnaya, M. A. (2002). Psikhologiya intellekta: Paradoksy issledovaniya[Psychology of intelligence: Paradoxes of research] (2nd ed., rev. and exp.).
13. Kretov, A. (2026). Chto takoe iskusstvennyi intellekt (ne neiroseti) i kakie u nego est’ vidy [What is artificial intelligence (not neural networks) and what types exist]. Habr. https://habr.com/ru/articles/1025094/
14. Maeda, T. (2024). Misplaced capabilities: Evaluating the risks of anthropomorphism in human-AI interactions. Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, 7(2), 35-36. https://doi.org/10.1609/aies.v7i2.31903
15. Menovshchikov, V. Yu. (2025). Iskusstvennyi intellekt v psikhologicheskom konsul’tirovanii i psikhoterapii (obzor otechestvennykh publikatsii) [Artificial intelligence in psychological counseling and psychotherapy (review of Russian publications)]. Psikhologicheskoe Konsul’tirovanie Onlain, (2), 14-31.
16. Nass, C., & Moon, Y. (2000). Machines and mindlessness: Social responses to computers. Journal of Social Issues, 56(1), 81-103. https://doi.org/10.1111/0022-4537.00153
17. Nass, C., Steuer, J., & Tauber, E. R. (1994). Computers are social actors. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems(pp. 72-78).  https://doi.org/10.1145/191666.191703
18. Panarin, A. S. (2010). Ratsionalizatsii protsess [Rationalization process]. In Novaya filosofskaya entsiklopediya(Vol. 4). Mysl’. https://iphlib.ru/library/collection/newphilenc/document/HASHe582c4644b9930e4c8eae5
19. Patrusheva, A., & Pavlova, A. (2025). Vtoroi razum: kak razvivaetsya iskusstvennyi intellekt i chto ego zhdet v budushchem[The second mind: How artificial intelligence is developing and what awaits it in the future]. Yandeks Praktikum. https://practicum.yandex.ru/blog/chto-takoe-iskusstvennyi-intellekt/
20. Porus, V. N. (2010). Ratsional’nost’ [Rationality]. In Novaya filosofskaya entsiklopediya(Vol. 4). Mysl’. https://iphlib.ru/library/collection/newphilenc/document/HASH6743a4d47bb13ceacfee67
21. R Core Team. (2026). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://doi.org/10.32614/R.manuals
22. (2026). Iskusstvennyi intellekt[Artificial intelligence]. https://sber.pro/publication/iskusstvennii-intellekt/
23. Sufyan, N. S., Fadhel, F. H., Alkhathami, S. S., & Mukhadi, J. Y. A. (2024). Artificial intelligence and social intelligence: Preliminary comparison study between AI models and psychologists. Frontiers in Psychology, 15, 1353022. https://doi.org/10.3389/fpsyg.2024.1353022
24. Sun, X., Hong, Q., Zhang, M., Li, Y., Chen, T., Huang, Z., Liang, G., Tang, W., Xu, S., Ni, X., Pang, J., Wan, P., & Long, E. (2026). Model confrontation and collaboration: A debate intelligence framework for enhancing medical reasoning in large language models. Cell Reports Medicine, 7(1), 102547. https://doi.org/10.1016/j.xcrm.2025.102547
25. Sundar, S. S. (2020). Rise of machine agency: A framework for studying the psychology of human-AI interaction (HAII). Journal of Computer-Mediated Communication, 25(1), 74-88. https://doi.org/10.1093/jcmc/zmz026
26. Suvorov, O. V. (2010). Intellekt [Intellect]. In Novaya filosofskaya entsiklopediya(Vol. 4). Mysl’. https://iphlib.ru/library/collection/newphilenc/document/HASH523b26975f21f2447bc452
27. Taylor, J. E. T., & Taylor, G. W. (2021). Artificial cognition: How experimental psychology can help generate explainable artificial intelligence. Psychonomic Bulletin & Review, 28, 454-475. https://doi.org/10.3758/s13423-020-01825-5
28. Ushakov, D. V. (2023). Artificial intelligence as a research instrument in psychology. Sibirskiy Psikhologicheskiy Zhurnal Siberian Journal of Psychology, 90, 188-200. https://doi.org/10.17223/17267080/90/11
29. Ushakov, D. V. (2024). Artificial intelligence technologies in psychology. Eksperimental’naya psikhologiya = Experimental Psychology (Russia), 17(4), 182-189. https://doi.org/10.17759/exppsy.2024170412
30. (2024). Doverie k II[Trust in AI]. https://wciom.ru/analytical-reviews/analiticheskii-obzor/doverie-k-ii
31. (2025). Neiroseti: instrument, a ne magiya[Neural networks: A tool, not magic]. https://wciom.ru/analytical-reviews/analiticheskii-obzor/neiroseti-instrument-a-ne-magija
32. (2026). Indeks tsifrofobii 2026[Digital phobia index 2026]. https://wciom.ru/analytical-reviews/analiticheskii-obzor/indeks-cifrofobii-2026
33. Waytz, A., Cacioppo, J. T., & Epley, N. (2010). Who sees human? The stability and importance of individual differences in anthropomorphism. Perspectives on Psychological Science, 5(3), 219-232. https://doi.org/10.1177/1745691610369336
34. Zhao, J., Wu, M., Zhou, L., Wang, X., & Jia, J. (2022). Cognitive psychology-based artificial intelligence review. Frontiers in Neuroscience, 16, 1024316. https://doi.org/10.3389/fnins.2022.1024316