An evaluation of the responsibility of generative AI models presents unique challenges that require holistic and practical solutions. This paper introduces an enhanced version of the VERIFAI framework, which extends beyond classification models to assess generative language models as well in terms of ethics, explainability, privacy, and security. Unlike existing theoretical frameworks, VERIFAI provides an integrated, software-driven approach that automates evaluations, ensures reproducibility, and offers actionable insights. To demonstrate its capabilities, we conduct an evaluation of the generative language model Llama-3.2-1B using the Regard metric, which quantifies bias in text generation. Our findings highlight systematic biases in model outputs, reinforcing the need for structured Responsible AI assessments. This work underscores VERIFAI’s scalability, intuitive UI, and advanced analysis capabilities, positioning it as a practical tool for the responsible evaluation of AI models.

The rise of Artificial Intelligence (AI) streamlines social media content generation, with ChatGPT sparking a surge in generative AI models. This impact is crucial for businesses enhancing digital marketing and researchers studying AI’s role in this realm. Marketing agencies and business leaders are actively engaging in the AI race with tailored models. Amid the multitude of applications, discerning, understanding, and effectively integrating appropriate tools into business practices are increasingly complex tasks. This paper addresses two questions: first, the potential of ChatGPT’s current version to transform social media content creation for micro-businesses, and second, the initial observations and changes in user experience over an extended use period. To answer the questions, we used a case study approach within the framework of Experiential Learning Theory (ELT), coupled with contextual inquiry research to reduce bias. Data from 15 specific ChatGPT interactions highlight its capabilities in fostering creativity for resource-strained micro-enterprises. This tool proves valuable for businesses without a designated social media marketing team, allowing them to consistently produce high-quality content, aim higher, and alleviate the pressure of finding perfect ideas for scaling in a competitive marketplace. ChatGPT serves as an ally, enhancing human capabilities and offering a transformative solution for micro-enterprises in content creation and marketing, however, there are limitations and concerns to be addressed.

Our investigation seeks to enhance the understanding of responsible artificial intelligence. The EU is deeply engaged in discussions concerning AI trustworthiness and has released several relevant documents. It’s crucial to remember that while AI offers immense benefits, it also poses risks, necessitating global oversight. Moreover, there’s a need for a framework that helps enterprises align their AI development with these international standards. This research will aid both policymakers and AI developers in anticipating future challenges and prioritizing their efforts. In our study, we delve into the essence of responsible AI and, to our understanding, introduce a comprehensive definition of the term. Through a thorough literature review, we pinpoint the prevailing trends surrounding responsible AI. Using insights from our analysis, we’ve also deliberated on a prospective framework for responsible AI. Our findings emphasize that human-centeredness should prioritized. This entails adopting AI techniques that prioritize ethical considerations, explainability of models, and aspects like privacy, security, and trustworthiness.

The “free” business model prevails in mobile apps available through the major channels, hinting at the possibility that users “pay” for the use of the mobile apps by sharing their private data with the developers and platform providers. Several types of personal data and permissions of mobile applications were analyzed. We examined 636 apps in several categories, such as medical, health & fitness, business, finance, and entertainment. The types of personal data being requested by the apps were collected from their privacy policies and the list of permissions was scraped from the platform’s store. We implemented a privacy policy word processing algorithm, the purpose of which was to gain a better insight into the types of data collected. Using the algorithm results, we also performed statistical analyses, based on which we found, expectedly, that free mobile applications collect more data than paid ones. However, there are discrepancies between the permissions we obtained from the privacy policy texts and those stated on the Google Play and Apple App Store websites. More permission requirements emerged from the privacy policy texts than were shown on corresponding app stores, which is a worrying result.

Privacy is a fundamental human right and is widely end extensively protected in the western industrialized world. The recent advances in technologies, especially in the use of applications developed and designed for mobile devices, have led to the rise of its abuse on one hand and a higher awareness of the importance of privacy on the other side. Legal texts protecting privacy have attempted to rectify some of the problems, but the ecosystem giants and mobile apps developers adapted. In this paper, we analyze which data mobile apps developers are collecting. We have focused on a sample of apps in the medical and health field. The research was done using collocations analysis. A relationship between a base word and its collocative partners was sought. The initial visual results have led us to more detailed studies that unveiled some worrying patterns. Namely, applications are collect data about the users and their activities, but also about their family members, medical diagnoses, treatments, and alike, going well beyond the “need to function” / functionality threshold.

Respect for privacy is not a modern phenomenon as it has been around for centuries. Recent advances in technologies led to the rise of awareness of the importance of privacy, and to the development of principles for privacy protection to guide the engineering of information systems on one side, and on using the principles to draft legal texts protecting privacy on the other side. In this paper, we analyze how respect for privacy has been implemented in GDPR by automated comparison of the similarity of GDPR’s articles and the text of seven principles of Privacy by Design. We have compared the specific text of GDPR’s first 50 core privacy-protecting articles and the GDPR’s remaining provisions to establish independent supervisory authorities. The first half is observing the privacy by design principles, each of them considerably more than the second half. Our findings show that automated similarity comparison can highlight portions of legal texts where principles were observed. The results can support drafting legal texts to check whether important legal (or other) principles were adequately addressed.

There is a large number of NoSQL data systems which can be classified into four different types of NoSQL databases. However, there are no generally well-known and established software quality frameworks that help information system developers decide which database or system is the most useful in their case. We present a comparison and definition of NoSQL systems according to selected quality attributes based on literature review, and an overview and evaluation of how different NoSQL solutions meet the most important quality criteria. Further contribution of this work is definition of a utility function which can, together with the evaluation, help developers, architects and software engineers to understand different quality attributes and how they reflect in selected NoSQL products, and how a certain NoSQL database system would help solve their particular problem, and based on utility function they can actually select the most appropriate solution.

It is very difficult to make a choice between star and snowflake data warehouse schema. The topic is part of the broader dilemma in the data warehousing community: which approach to use, Kimball's or Inmon's. There are advocates for each approach, with fierce “war” still going on. However, very few empirical studies exist giving either side an advantage; in the past, the approaches were being selected based on the organizational, resource or goal-specific parameters. The goal of this case study was to examine which implementation of data warehouse will yield better results in the observed scenario – a data warehouse for monitoring of energy consumption in public buildings, from perspective of the ETL process. We implemented two versions of DW, one based on star and the other on the snowflake schema model, and measured the performance of the ETL process. Our goal was to find out if there is a difference in duration between two implementations, and if the difference exists, how it changes with increase of data in operational database. Series of tests were conducted to evaluate implemented solutions. Statistical analysis showed that, for the observed scenarios, implementation based on snowflake schema performs better: the ETL execution time is shorter and the size of DW is smaller. An important observation is that the target data warehouse size is linearly dependent on the amount of the operational data.

The power law is predominantly describing the ways humans learn, especially in psychophysics, in skill acquisition, and in retention. Yet a few researchers claim that this law is applicable only on the aggregate level and that exponential law should be considered when describing a single learning process. The question which law should be used on aggregate or single learner level has not yet been answered in the artificial learning community. This work is shedding some light towards the answers. We conducted an experiment with three artificial learners using 109 training cases. The statistical tests have shown that power law and exponential law are describing the learning curves equally well. However, in quite many cases neither of laws is applicable. Additionally, there are significant differences among artificial learners.

Background: A general, restrictions-free theory on performance of arbitrary artificial learners has not been developed yet. Empirically, not much research has been per-formed on the question of an appropriate description of artificial learner's performance.

Background: Measuring the performance of a classifier is crucial when trying to find the best machine-learning algorithm with optimal parameters. Multiple methods are available in this regard and the more common is the k-fold cross-validation. Other similar methods are the bootstrap method and the k-fold repeated cross-validation.