EXPRESS: Marketing AI Superiority: Risks of Using the ‘Better-Than-Human’ Framing

(Potential) users of artificial intelligence (AI) products (e.g., smart fridges, voice assistants) are most often not fully informed about the data security and data privacy (DSDP) of these AI products as such information is not available in an easily accessible format. Here, I argue that it is essential to provide (potential) users with such information presented in an easily accessible and quick-to-process format as this could help instill appropriate levels of (dis-)trust and thus support (trustworthy) AI acceptance and adoption. Testing the effect of AI trustworthiness labels, I presented participants with hypothetical AI products that were versus were not paired with a graphical or text-based label indicating a low to high DSDP level. Trust, AI anxiety, as well as the monetary value attributed to AI products scaled with an AI trustworthiness label’s DSDP level. Importantly, participants’ ratings of unlabeled AI products corresponded to their perceptions of AI products labeled with an intermediate DSDP level. This apparent bias towards intermediate DSDP judgements in the absence of information on AI products underscores the relevance of explicitly communicating AI trustworthiness to (potential) users. Interestingly, differences in trust for AI products with a high as compared to intermediate DSDP level further increased with more positive attitudes towards AI technology.

Artificial intelligence is widely used in healthcare. World leading scientific schools are experimenting with various machine learning models and trying to implement them into medical practice. In various fields of medicine, there are good solutions already exists that allow artificial intelligence products use as medical devices for making medical decisions. For example, artificial intelligence products received a special impetus to this during COVID-19 in radiology. For high-quality artificial intelligence training, high-quality datasets are needed. That’s why as in general artificial intelligence, the transition from a model-centric approach to a data-centric approach is gradually moving. The data quality for neural networks training is becoming more and more acute. There are many problems associated with the statistically reliable medical datasets collection. The article considers an innovative solution of recent years - synthetic datasets, which can solve many dataset quality problems for artificial intelligence training. Key factors affecting the complexity of medical data mining are analyzed and examples of synthetic datasets can reduce artificial intelligence development time, balance data and improve the quality of artificial intelligence products are presented.KeywordsArtificial intelligenceSynthetic datasetsMedical artificial intelligence products

Objectives This study explored the familiarity, perceptions and confidence of Australian radiology clinicians involved in reading screening mammograms, regarding artificial intelligence (AI) applications in breast cancer detection. Methods Sixty-five radiologists, breast physicians and radiology trainees participated in an online survey that consisted of 23 multiple choice questions asking about their experience and familiarity with AI products. Furthermore, the survey asked about their confidence in using AI outputs and their preference for AI modes applied in a breast screening context. Participants' responses to questions were compared using Pearson's χ 2 test. Bonferroni-adjusted significance tests were used for pairwise comparisons. Results Fifty-five percent of respondents had experience with AI in their workplaces, with automatic density measurement powered by machine learning being the most familiar AI product (69.4%). The top AI outputs with the highest ranks of perceived confidence were 'Displaying suspicious areas on mammograms with the percentage of cancer possibility' (67.8%) and 'Automatic mammogram classification (normal, benign, cancer, uncertain)' (64.6%). Radiology and breast physicians preferred using AI as second-reader mode (75.4% saying 'somewhat happy' to 'extremely happy') over triage (47.7%), pre-screening and first-reader modes (both with 26.2%) (P < 0.001). Conclusion The majority of screen readers expressed increased confidence in utilising AI for highlighting suspicious areas on mammograms and for automatically classifying mammograms. They considered AI as an optimal second-reader mode being the most ideal use in a screening program. The findings provide valuable insights into the familiarities and expectations of radiologists and breast clinicians for the AI products that can enhance the effectiveness of the breast cancer screening programs, benefitting both healthcare professionals and patients alike.

The digital age provides, on the one hand, extraordinary opportunities for modern information technologies, including artificial intelligence (AI) products; on the other hand, it creates new problems for humanity, for example, of a legal and ethical nature. In addition, AI products can be highly effective and attractive from a commercial point of view; and from the point of view of humanity as a whole, they can signify a huge problem. Thus, there is a need to balance the value of these products, taking into account their multiaspectual nature. The purpose of this study is to develop a conceptual approach to assessing and balancing the value of AI products. Balancing involves assessing the value of AI products, comparing it to some standards, and adjusting it based on the identified inconsistencies. It is found that balancing the value of AI products should take into account three areas of view and evaluation – individual, business, society (humanity). Such an integral view will provide coverage of the main levels that are affected by AI products. A method for assessing the value of AI products has been proposed, which takes into account the multiaspectual nature of product consideration for each area (individual, business, society) and the priority of each aspect. It is also proposed that in the future, it is artificial intelligence that should provide this assessment of the value of AI products, taking into account a significant amount of necessary information. A conceptual scheme has been formed that links the increase in the level of development of AI products with the level of their responsibility, especially for those products that are designed to evaluate and determine the necessary areas of harmonization and balancing of value. This will ensure the development and commercial use of AI products with risk minimization from various perspectives. It is evident that these results are only a conceptual approach, on the basis of which it is possible to develop specific algorithms for evaluating and balancing the value of AI products.

Currently, artificial intelligence is entering a rapidly developing era. Artificial intelligence technology is growing rapidly. Artificial intelligence technology can already be applied to various fields of social life, and artificial intelligence products with artificial intelligence technology as the core are abundant. Artificial intelligence products have been integrated into the lives of ordinary people. Due to the autonomy of artificial intelligence products, they may work in an unmanned state at certain times, resulting in infringement liability that may cause personal or property damage to others during the operation of artificial intelligence products. There are two ways to pursue accountability: one is to directly pursue the infringement liability of artificial intelligence products, and the specific subject who caused the infringement of artificial intelligence products shall bear the infringement liability; another approach is to pursue the responsibility of the producer and seller of artificial intelligence products for their subsequent observation obligations. The reason is that the producer and seller of artificial intelligence products fail to fulfill their subsequent observation obligations, resulting in the failure to detect defects in the artificial intelligence products in a timely manner and resulting in subsequent observation obligations. The responsible parties are the producers and sellers of artificial intelligence products.

Drawing on the analytic of the “colonial matrix of power” developed by Aníbal Quijano within the Latin American modernity/coloniality research program, this article theorises how a system of coloniality underpins the structuring logic of artificial intelligence (AI) systems. We develop a framework for critiquing the regimes of global labour exploitation and knowledge extraction that are rendered invisible through discourses of the purported universality and objectivity of AI. ​​Through bringing the political economy literature on AI production into conversation with scholarly work on decolonial AI and the modernity/coloniality research program, we advance three main arguments. First, the global economic and political power imbalances in AI production are inextricably linked to the continuities of historical colonialism, constituting the colonial supply chain of AI. Second, this is produced through an international division of digital labour that extracts value from majority world labour for the benefit of Western technology companies. Third, this perpetuates hegemonic knowledge production through Western values and knowledge that marginalises non-Western alternatives within AI’s production and limits the possibilities for decolonising AI. By locating the production of AI systems within the colonial matrix of power, we contribute to critical and decolonial literature on the legacies of colonialism in AI and the hierarchies of power and extraction that shape the development of AI today.

Several countries, including the UK, have approved commercial artificial intelligence (AI) products as medical devices to assist in the diagnosis of skin cancer. However, lack of robust real-world data has led to professional bodies recommending against their use outside research environments. The aim of this study was to assess the performance of an internationally proven deep neural network as a screening tool in patients referred from primary care on the suspected skin cancer pathway. Dermoscopic images were obtained from patients attending a community lesion imaging clinic following referral from primary care on the suspected skin cancer pathway. Ultimate diagnosis was based on histology where available, face-to-face clinical opinion if no histology was obtained, and the opinion of two dermatology specialists with over 10 years’ experience for patients discharged directly from teletriage without histology or face-to-face review. Images were assessed by the neural network and binary categorization was performed. The algorithm was set to detect invasive malignancies and premalignant conditions (actinic keratosis, Bowen disease, keratoacanthoma) and was trained to dismiss benign naevi, benign keratotic lesions (seborrhoeic keratosis, lichen planus-like keratosis, solar lentigines), dermatofibromata and benign vascular lesions. In total, 333 of 382 lesions from consecutive patients who gave research consent were included. Reasons for exclusion were pen artefact on dermoscopic image (26), lesion resolved prior to imaging clinic (eight) inappropriate site (six), poor image quality (five) and other (four). Ultimate diagnoses were basal cell carcinoma (62), squamous cell carcinoma (19), melanoma (11), other malignancy (one), benign keratotic (80), actinic keratosis (61), Bowen disease (20), benign naevus (17), benign vascular (nine), keratoacanthoma (eight) and other benign (45). Overall, 81 of 333 lesions (24.3%) were classified by the algorithm as benign and 252 (75.7%) were flagged as possibly malignant or premalignant. Sensitivity for cancer was 100% (93 of 93 detected). Sensitivity for premalignant conditions was 97% (86 of 89 detected). Specificity for all patients was 51.7% (78 of 151 benign lesions correctly classified). Specificity when removing the ‘other benign’ category, which is in the algorithm’s exclusion criteria and not part of its training, was 65.1% (71 of 109 lesions accurately categorized). Two actinic keratoses and one cases of Bowen disease were incorrectly classified as benign. The algorithm performed extremely well on this real-life case series and detected 100% of invasive malignancies. AI screening of suspected skin cancer referrals is likely to play a significant part in addressing future National Health Service capacity–demand mismatch and would have reduced the caseload requiring specialist opinion by 24.3% in this series. There is urgent need in the UK for further transparent clinical trials and open competition between AI products on large real-world datasets. Further stratification of benign and premalignant conditions may lead to greater benefit than binary classification.

Artificial Intelligence and the Pathologist: Future Frenemies?

Artificial General Intelligence is the idea that someday an hypothetical agent will arise from artificial intelligence (AI) progresses, and will surpass by far the brightest and most gifted human minds. This idea has been around since the early development of AI. Since then, scenarios on how such AI may behave towards humans have been the subject of many fictional and research works. This paper analyzes the current state of artificial intelligence progresses, and how the current AI race with the ever faster release of impressive new AI methods (that can deceive humans, outperform them at tasks we thought impossible to tackle by AI a mere decade ago, and that disrupt the job market) have raised concerns that Artificial General Intelligence (AGI) might be coming faster that we thought. In particular, we focus on 3 specific families of modern AIs to develop the idea that deep neural networks, which are the current backbone of nearly all artificial intelligence methods, are poor candidates for any AGI to arise due to their many limitations, and therefore that any threat coming from the recent AI race does not lie in AGI but in the limitations, uses, and lack of regulations of our current models and algorithms. This article appears in the AI &amp; Society track.

Dear Editors, I was very interested in your editorial [1]. Technological advancements are so rapid that humans are unable to keep up. And we eagerly await the outcomes of technological advancement. The study of artificial intelligence is one of the topics that has recently gained popularity. If someone had predicted a hundred years ago, "A technology will come that will detect the commands you receive, act accordingly, and write scientific articles," we would laugh at her. Many artificial intelligence dreams that appear weird to us now, I believe, will come true shortly. The Generative Pre-Trained Transformer 3 (GPT3) was developed a few years ago, in line with improvements in artificial intelligence. With this model, artificial intelligence was used to generate content that resembled human-written documents. A more advanced version of ChatGPT was produced a few years later. In March 2023, the most recent GPT4 version was launched. Errors in article writing have been reduced using this and comparable applications such as artificial intelligence. In fact, it has become nearly impossible to detect the difference between publications written by scientists and articles written by artificial intelligence. Some journals have begun to accept artificial intelligence apps as co-authors [2]. Article authoring is made considerably easier by artificial intelligence. In terms of time savings, using artificial intelligence products such as ChatGPT and GPT 4 may make sense. However, there are some risks associated with this circumstance. For example, because disseminating personal data is a felony, authors may feel compelled to conceal some information in order to preserve personal data. Artificial intelligence programs can readily perpetrate crimes involving personal data by releasing secret facts. This situation presents numerous legal issues. Furthermore, the language to be used in article writing may alter depending on the topic of study. In this instance, it is vital to go check the words chosen by artificial intelligence programs in the article. As a result, my recommendation is to use tools such as ChatGPT, GPT 4, or similar basic jobs once the user has written the article themselves. Leaving all of the work to these apps may cause more harm than good. Kind Regards

This paper investigates electromagnetic compatibility (EMC) problems in artificial intelligence (AI) products and summarize corresponding solutions. Concept and theories of EMC are introduced in Section II, analysis of EMC problems in AI products basing on three elements of EMC is given in Section III, solutions are presented in Section IV, and conclusions are given in the final Section.

This paper explains the ethical implications of artificial intelligence (AI) development towards achieving the level of artificial general intelligence (AGI) and analyzes the need for its social control. With the acceleration and intensification of AI growth and development, especially with the ongoing AI race, the transition from narrow AI to AGI becomes certain. The achievement of generative AI, which climaxes with chatbots (such as ChatGPT and others), transforms AI into a machine capable of creation. Although this AI application still appears relatively limited by algorithms, its learning ability is remarkable, and continuous advancements and the launch of increasingly sophisticated versions bring it ever closer to the AGI model. Each day brings us closer to that moment, which will signify AI’s transition from narrow AI to AGI. Unlike narrow AI, AGI deeply delves into the realm of ethics, and interpersonal and social relationships. Regulating AI-related policy and legally controlling AI represents one of the most serious and complex issues. In recent years, the community, led by corporate executives developing AI, prominent experts, researchers, scientists, writers, and other stakeholders, has made significant steps towards raising public awareness of the risks posed by advanced AI and making decisions, initiatives, and measures for monitoring, analyzing, and socially controlling the use of AI.

Follow-up of Artificial Intelligence Development and its Controlled Contribution to the Article: Step to the Authorship?

In South Korea, artificial intelligence (AI)-related personal information protection has long been an issue, and the Lee-Luda (also referred to as ‘Iruda’) incident drove the development of specific policies and guidelines. The Lee-Luda case is fairly important for AI governance in Korea, since the Personal Information Protection Commission (PIPC) decision was the first to make a judgment on the Personal Information Protection Act (PIPA) violations as a result of development and operation of AI. Following the Lee-Luda case, Korean society became more aware of the risks posed by privacy infringement in the age of intelligent information technology. Lawmakers began to establish legal frameworks for the protection of personal information while developing AI products and services. This study reviews legal issues and suggests policy directions while highlighting the significance of the Lee-Luda case, which was an epochal turning point in the protection of personal information related to AI products and services in Korea. First, the authors review the PIPC’s decision on Scatter Lab’s violation of the PIPA, analyse the legal issues in the Lee-Luda case, and derive the significance and limitations of the decision. Next, the authors suggest that it is desirable for AI governance to move towards self-regulation in a manner that ensures the autonomy of AI developers and operators; they also review the recent trends in the laws and policies on data protection that have been continually developed since the Lee-Luda case, especially with a focus on the AI Personal Information Protection Self-Checklist (Self-Checklist).

The study examined the effect of artificial intelligence (AI) disclosure on the financial performance of banks in Nigeria from 2020 to 2024. Specifically, the study sought to: determine the effect of AI applications disclosures on financial performance, investigate the effect of AI products disclosures on financial performance, and examine the effect of AI services disclosures on financial performance within the Nigerian banking sector. A review of relevant conceptual, theoretical, and empirical literature was conducted, and the study was anchored on agency theory. The research adopted an ex post facto design. The population comprised all listed commercial banks in Nigeria, from which a sample of six banks was selected using a convenience sampling technique. Data were sourced from the annual reports of the sampled banks. Descriptive statistics, correlation analysis, and multiple regression analysis were employed to analyze the data. The findings indicate that AI applications disclosure had a significant positive effect on the financial performance of banks. Similarly, AI products disclosure also exhibited a significant positive effect on financial performance. However, AI services disclosure was found to have no significant effect on financial performance. The study concludes that overall, AI disclosure positively influences the financial performance of banks in Nigeria. It recommends that policies promoting the disclosure of AI applications and products such as mandatory reporting on AI initiatives should be encouraged, as these disclosures have been shown to enhance banks’ financial performance.