Through our research, we found that many existing drone systems do not clearly solve the problems related to thermal detection in disaster areas, and they often fail in providing reliable real-time human detection. These issues are important to address because search and rescue operations need to be more efficient, time-saving, and economic. Since UAVs are extremely useful in locations where humans cannot reach easily, improving their performance in such missions becomes essential for future emergency response systems. In this work, we propose a drone system designed to detect humans more effectively by using audio, video, and thermal inputs. Our approach focuses on integrating GPS, computer vision, and artificial intelligence to automate the drone’s operation and support real-time data processing. The drone captures photos, videos, audio, or thermal images, and the AI model confirms whether a human is present and potentially in need. Once detection is verified, the system immediately processes and sends the information to rescue teams in real time. It can also hold its position or circle around the target until responders arrive. The implications of this system show that faster alerts, improved detection accuracy, and economic design can significantly support rescue teams and help save more lives during emergencies.

Background: Multimodal large language models (MLLMs) integrating multiple AI systems and unimodal large language models (LLMs) represent distinct approaches to clinical decision support. Their comparative performance against human clinical experts in complex cardiovascular emergencies remains inadequately characterized. Objective: To compare the performance of a combined MLLM system (GPT-4V + Med-PaLM 2 + BioGPT), a unimodal LLM (ChatGPT-5.2), and human physicians from multiple centers (radiologists, emergency medicine specialists, cardiovascular surgeons) on aortic dissection clinical questions across diagnosis, treatment, and complication management domains. Methods: This multicenter cross-sectional study was conducted across five tertiary care centers in Turkey (Elazığ, Ankara, Antalya). A total of 25 validated multiple-choice questions were categorized into three domains: diagnosis (n = 8), treatment (n = 9), and complication management (n = 8). Questions were administered to the MLLM, ChatGPT-5.2 (Unimodal), and nine physicians from five centers: radiologists (n = 3), emergency medicine specialists (n = 3), and cardiovascular surgeons (n = 3). Statistical comparisons utilized chi-square tests. Results: Overall accuracy was 92.0% for the MLLM and 96.0% for ChatGPT-5.2 (Unimodal). Among human physicians, cardiovascular surgeons achieved 96.0%, radiologists 92.0%, and emergency medicine specialists 89.3%. The MLLM excelled in diagnosis (100%) but showed lower performance in treatment (88.9%) and complication management (87.5%). No significant differences were observed between AI models and human physician groups (all p > 0.05). Conclusions: Both the MLLM and unimodal ChatGPT-5.2 demonstrated performance within the range of human clinical experts in this controlled assessment of aortic dissection scenarios, though definitive conclusions regarding equivalence require larger-scale validation. These findings support further investigation of complementary roles for different AI architectures in clinical decision support.

This study aimed to develop a consensus-based set of patient questions on dental implant failure and to compare the clarity, quality, accuracy, reliability, and readability of responses generated by 4 widely used AI chatbots: ChatGPT-4, DeepSeek-R1, Microsoft Copilot, and Google Gemini. Twenty-three expert-validated questions were derived from the EAO 2021 and ICOI Pisa Consensus reports and independently submitted to each AI model under standardized, non-personalized conditions. Responses were assessed using CLEAR criteria, mGQS, a 5-point accuracy scale, the first 8 DISCERN items, and Flesch-based readability indices. Nonparametric tests were used for intermodel comparisons. AI models demonstrated significant variability in performance. Gemini achieved the highest accuracy (P<0.001), whereas ChatGPT-4 exhibited the highest reliability based on DISCERN scores. Copilot generated the most structurally fluent responses, whereas DeepSeek-R1 offered the best readability. Although CLEAR and mGQS scores were high across all systems, readability and linguistic complexity varied markedly. Accuracy, clarity, and reliability were strongly correlated, whereas readability displayed the expected inverse association with grade-level demand. AI chatbots hold potential as adjunct tools for patient education on implant failure; however, their performance characteristics differ substantially. Gemini excels in accuracy, ChatGPT-4 in reliability, Copilot in fluency, and DeepSeek-R1 in readability. Model-specific guidance and continued refinement are needed to enhance the clinical usefulness and accessibility of AI-generated patient information.

What is the fastest Artificial Intelligence Large Language Model (AI LLM) for generating quantum operations? To answer this, we present the first benchmarking study comparing popular and publicly available AI models tasked with creating quantum gate designs. The Wolfram Mathematica framework was used to interface with the six AI LLMs, including Google Gemini 2.0 Flash, Anthropic Claude 3 Haiku, WolframLLM Notebook Assistant For Mathematica V14.3.0.0, OpenAI ChatGPT Omni 4 Mini, Google Gemma 3 4b 1t, and DeepSeek Chat V3. Our novel study found the following: (1) Gemini 2.0 Flash is overall the fastest AI LLM of the models tested in producing average quantum gate designs at 2.66101 s, factoring in the “thinking” execution time and ServiceConnect network latencies. (2) On average, four out of the ten quantum operations that the six LLMs produced compiled in Python (40.8% success rate). (3) Quantum operations averaged approximately 21–45 Lines of Code (omitting nonsensical outliers). (4) DeepSeek Chat V3 produced the shortest code with an average of 21.6 lines. This comparison evaluates the time taken by each AI LLM platform to generate quantum operations (including ServiceConnect networking times). These findings highlight a promising horizon where publicly available Large Language Models can become fast collaborators with quantum computers, enabling rapid quantum gate synthesis and paving the way for greater interoperability between two remarkable and cutting-edge technologies.

Risk management in large-scale construction projects is a critical yet complex process influenced by financial, safety, environmental, scheduling, and regulatory uncertainties. Effective risk management contributes directly to project optimization by minimizing disruptions, controlling costs, and enhancing decision-making efficiency. Early identification and mitigation of risks allow resources to be allocated where they have the greatest effect, thereby optimizing overall project outcomes. However, conventional methods such as expert judgment and probabilistic modeling often struggle to process extensive datasets and complex interdependencies among risk factors. This study explores the potential of an AI-based framework for risk identification, utilizing artificial intelligence to analyze project documentation and generate a preliminary set of identified risks. The proposed methodology is implemented on the ‘Trg pravde’ judicial infrastructure project in Zagreb, Croatia, applying AI models (GPT-5, Gemini 2.5, Sonnet 4.5) to identify phase-specific risks throughout the project lifecycle. The approach aims to improve the efficiency of risk identification, reduce human bias, and align with established project management methodologies such as PM2. Initial findings suggest that the use of AI may broaden the range of identified risks and support more structured risk analysis, indicating its potential value as a complementary tool in risk management processes. However, human expertise remains crucial for prioritization, contextual interpretation, and mitigation. The study demonstrates that AI augments, rather than replaces, traditional risk management practices, enabling more proactive and data-driven decision-making in construction projects.

This quality improvement study assesses the vulnerability of leading commercial large language models to invisible text injection manipulation in simulated medical peer review.

The pressure of climate change on water resources has made the development of reliable hydrological models increasingly important, especially for data-scarce regions. However, due to the limited availability of ground-based observations, it considerably affects the accuracy of models developed using these inputs. This also limits the ability to investigate future hydrological behavior. Satellite-based data sources have emerged as an alternative to address this challenge and have received significant attention. However, the transferability of these datasets across different model classes has not been widely explored. This paper evaluates the transferability of satellite-derived inputs to eleven types of models, including process-based (SWAT), data-driven methods (XGBoost and WGAN), and hybrid model structures that utilize SWAT outputs with AI models. SHAP has been applied to overcome the black-box limitations of AI models and gain insights into fundamental hydrometeorological processes. In addition, uncertainty analysis was performed for all models, enabling a more comprehensive evaluation of performance. The results indicate that hybrid models using SWAT combined with WGAN can achieve better predictive accuracy than the SWAT model based on ground observation. While the baseline SWAT model achieved satisfactory performance during the validation period (NSE ≈ 0.86, KGE ≈ 0.80), the hybrid SWAT + WGAN framework improved simulation skill, reaching NSE ≈ 0.90 and KGE ≈ 0.89 during validation. Models forced with satellite-derived meteorological inputs additionally performed as well as those forced using station-based observations, validating the feasibility of using satellite products as alternative data sources. The future hydrological status of the basin was assessed based on the best-performing hybrid model and CMIP6 climate projections, showing a clear drought signal in the flows and long-term reductions in average flows reaching up to 58%. Overall, the findings indicate that the proposed framework provides a consistent approach for data-scarce basins. Future applications may benefit from integrating spatio-temporal learning frameworks and ensemble-based uncertainty quantification to enhance robustness under changing climate conditions.

Hierarchical Federated Learning (HFL) alleviates the trade-off between communication overhead and privacy protection in mobile scenarios via multi-level aggregation and mobility consideration. However, its idealized convex loss assumption and full-dimension parameter upload deviate from real-world non-convex tasks and edge channel constraints, causing excessive energy consumption, high communication cost, and compromised convergence that hinder practical deployment. To address these issues in mobile/UAV networks, this paper proposes an energy-efficient optimization scheme for HFL under non-convex loss, integrating a dynamically adjustable partial-dimension model upload mechanism. By screening key update dimensions, the scheme reduces uploaded data volume. We construct a total energy minimization model that incorporates communication/computation energy formulas related to upload dimensions and introduces an attendance rate constraint to guarantee learning performance. Using Lyapunov optimization, the long-term optimization problem is transformed into single-round solvable subproblems, with a step-by-step strategy balancing minimal energy consumption and model accuracy. Simulation results show that compared with the original HFL algorithm, our proposed scheme achieves significant energy reduction while maintaining high test accuracy, verifying the positive impact of mobility on system performance.

Sepioids are an evolutionarily successful group of modern ten-armed cephalopods (Decabrachia) of high biodiversity, providing a large amount of biomass in present-day oceans. They include the internally shelled order Sepiida (cuttlefish) and the soft-bodied order Sepiolida (bobtail squid). The phylogenetic position and evolutionary history of these orders are, however, so far poorly understood due to the patchy fossil record of the Decabrachia. Here we report Uluciala rotundata gen. et sp. nov. from the upper Campanian to upper Maastrichtian (~74-67 Ma, Upper Cretaceous), South Dakota, which shows an intermediate morphology between Sepiida and Sepiolida. This discovery was facilitated by a new approach in palaeontology, the Digital fossil-mining method incorporating a zero-shot learning AI model. Uluciala rotundata demonstrates a close relationship between the two sepioid orders, which has previously been interpreted controversially. Our findings indicate that sepioids experienced an early phase of radiation in the later part of the Late Cretaceous.

Context: Bringing voice-controlled interfaces into Internet of Things (IoT) systems has created fresh opportunities for smart environments. However, existing voice assistants often struggle with non-standardized languages, especially Arabic dialects. Objective: This research paper explores the challenges and potential of integrating five Arabic dialect variants, namely Modern Standard Arabic (MSA), known as Fusha (الفصحى), Egyptian, Levantine, Gulf, and Algerian dialects, into AI-driven IoT systems. Methods: For each dialect, a comparative simulation was performed using two AI models: a baseline model and a dialect-aware model. Key simulated metrics included automatic speech recognition (ASR) accuracy, intention recognition, task success rate, and system response time. Results: The results consistently show that the dialect-aware model outperforms the baseline model in all metrics. It provides higher ASR and intention recognition accuracy, improved task success rates, and faster response times, especially for regional dialects. The Algerian dialect, while still challenging, benefited significantly from the dialect-aware adaptations of the improved model. These results highlight the potential of dialect-aware AI to close the performance gap caused by linguistic variation and code-switching. Conclusion: This study highlights the importance of considering linguistic diversity when developing accessible, culturally appropriate IoT interfaces that ensure a more inclusive and natural user interaction.

This study explores how incorporating artificial intelligence improves institutional resilience and overcomes the rigidity of conventional, data-based methods to alter financial risk management. To find patterns in AI applications, resilience theory, and integration pathways, a qualitative systematic literature review was carried out utilizing theme synthesis in accordance with PRISMA peer-reviewed protocols. Findings show that AI techniques, machine learning for tail-risk detection, deep learning for high-frequency forecasting, and explainable AI for transparent decisions, yield up to 28% reductions in forecasting errors and halve recovery times during crises. The hybrid CNN Transformer architectures and transformer-based NLP models significantly enhance predictive accuracy and forward-looking insights. The study suggests financial institutions adopt integrated AI frameworks, invest in data quality and human–AI collaboration, and implement principle-based governance to balance innovation with fairness and stability. Limitations include reliance on published literature and limited representation of emerging AI models, warranting future longitudinal and context-specific empirical research.

Accurately tracking the global distribution of precipitation is essential for both research and operational meteorology. Satellite observations remain the only means of achieving consistent, global precipitation monitoring. While machine learning has long been applied to satellite-based precipitation retrieval, the absence of a standardized benchmark dataset has hindered fair comparisons between methods. To address this, the International Precipitation Working Group has developed SatRain, the first AI benchmark dataset for satellite-based detection and estimation of rain. SatRain integrates multi-sensor satellite observations from the primary platforms used in precipitation remote sensing with high-quality reference precipitation estimates derived from gauge-corrected ground-based radar composites over the conterminous United States. It offers a standardized evaluation protocol and out-of-distribution testing data from Asia and Europe to enable robust and reproducible comparisons across machine learning approaches. In addition to algorithm evaluation, the diversity of sensors and inclusion of time-resolved geostationary observations make SatRain a valuable foundation for developing next-generation AI models to deliver more accurate global precipitation estimates.

Background: Magnetic Resonance Imaging (MRI) has emerged as a quintessential tool in the assessment and control of brain tumors due to its non-invasive nature and advanced soft-tissue assessment. This evaluation examines the modern-day applications and destiny possibilities of MRI in brain tumor evaluation. Traditional structural imaging, inclusive of T1- and T2-weighted imaging, plays a critical position in figuring out tumor place, size, and related edema, at the same time as advanced strategies offer deeper insights into tumor biology. Functional imaging modalities inclusive of Diffusion-Weighted Imaging (DWI) and Perfusion-Weighted Imaging (PWI) permit differentiation of tumor grades, identity of remedy results, and evaluation of tumor cellularity and vascularity. Magnetic Resonance Spectroscopy (MRS) enables the analysis of tumor metabolism, providing treasured records on biomarkers including choline and lactate. Emerging techniques like Chemical Exchange Saturation Transfer (CEST) MRI are being advanced for extra precise molecular and metabolic characterization. Future advancements encompass the mixing of synthetic intelligence (AI) for computerized tumor detection, category, and prediction of healing responses. AI models blended with radiomic evaluation hold promise for customized treatment strategies. Intraoperative MRI has more desirable surgical effects by using allowing actual-time imaging, enhancing tumor resection accuracy, and retaining healthy tissue. Furthermore, extremely-high-area MRI (7T) gives extraordinary spatial and contrast decision, facilitating particular evaluation of tumor microenvironments. Despite its transformative effect, challenges continue to be, inclusive of excessive charges, accessibility problems, and interpretation variability. The aim of this review is to explore the current and future advancements in MRI technology for brain tumor evaluation, highlighting the integration of AI, advanced imaging techniques, and intraoperative MRI to improve tumor detection, treatment planning, and surgical outcomes. It also addresses challenges such as cost, accessibility, and interpretation variability.

Background/Objectives: The aim of the present systematic review is to evaluate the performance of AI models for length of stay prediction. Methods: This SR was carried out in accordance with PRISMA 2020 and registered in PROSPERO database (CRD420251039985). Using the PICOS framework, we formulated the following research question: “Can artificial intelligence models accurately predict hospital length of stay (LOS) in patients undergoing head and neck (H&amp;N) cancer surgery?” We searched the Cochrane Library, Embase, PubMed, and Scopus, with additional gray literature identified through Google Scholar and ProQuest. Risk of bias (RoB) was assessed using the Prediction Model Risk of Bias Assessment Tool (PROBAST), and a narrative synthesis was performed to summarize qualitative findings. Results: Of 1304 identified articles, 5 met inclusion criteria, covering 5009 patients. All studies used supervised learning to predict LOS with different variables presenting stronger associations with increased hospital LOS. Age, race, ASA score, BMI, and comorbid factors like smoking and arterial hypertension were comon variables across studies but not always the ones most strongly associated with LOS. One study also predicted discharge to non-home facilities and prolonged LOS; only one applied data balancing. Model accuracies ranged from 0.63 to 0.84, and area under the receiver operator characteristics curve (AUROC) values from 0.66 to 0.80, suggesting moderate discriminative performance. All studies had a high risk of bias, though no applicability concerns were noted. Conclusions: AI models show potential for LOS prediction after H&amp;N cancer surgery; however, an elevated RoB and methodological shortcomings constrain the current evidence. Methodological improvements, external validation, and transparent reporting is essential to enhance reliability and generalizability, enabling integration into clinical decision-making.

Generative AI systems are increasingly relied on and are already actively reshaping how we think about privacy and data protection law. Models ingest and process vast amounts of personal and sensitive data, challenging assurances of compliance with legal frameworks like the General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA) with increasing intensity. Machine unlearning is an emerging tool in practitioners’ attempts to address these challenges: the act of selectively removing or suppressing specific data, such as personal data that a data subject requests be deleted, from AI models as means of complying with legal obligations or policy goals. This Article’s much-needed analysis of unlearning’s technicalities and uses builds on recent critical scholarship that examines unlearning’s limitations at the technical and policy level. It delves deeper into machine unlearning’s implications for privacy and data protection law by situating it within privacy law’s broader ecosystem and proposing actionable pathways for integrating unlearning into enforcement and policy. Specifically, this Article evaluates whether privacy laws’ legal, remedial, and normative aspirations can be reconciled with the technical realities of machine unlearning in generative AI systems. It also contributes to the privacy profession by proposing a framework for integrating machine unlearning into broader privacy-preserving interventions. In doing so, the Article positions machine unlearning as both a vital new tool as well as a site of contestation in the evolving landscape of privacy and AI governance while providing a forward-looking roadmap for aligning machine unlearning with privacy law’s goals.

The increasing reliance on automatic code generation integrated with Generative AI technology has raised new challenges for cybersecurity defense against code injection, insecure code templates, and adversarial manipulation of an AI model. These risks make developing advanced frameworks imperative to ensure secure, reliable, and privacy-preserving code generation processes. The paper presents a novel Hybrid Artificial Neural Network (ANN)-Interpretive Structural Modeling (ISM) Framework to alleviate the cybersecurity risks associated with the automatic code generation using Generative AI. The proposed framework integrates the predictive capability of ANN and structured analysis of ISM for the identification, evaluation, and treatment of common vulnerabilities and risks in automatic code generation. We first conduct a multivocal literature review (MLR) to identify cybersecurity risks and generative AI practices for addressing these risks in automatic code generation. Then we conduct a questionnaire survey to identify and validate the identified risks and practices. An expert panel review was then assigned for the process of ANN-ISM. The ANN model can predict potential security risks by learning from historical data and code generation patterns. ISM is used to (1) structure and visualize (2) relations between identified risks and mitigation approaches and (3) offer a combined, multi-layered risk management methodology. We then perform an in-depth examination of the framework with a case study of an AI-based code generation company. We further determine its practicality and usefulness in real-world settings. The case study results show that the framework efficiently handles the primary cybersecurity challenges, such as injection attacks, code quality, backdoors, and lack of input validation. The analysis characterizes the maturity of several mitigation practices and areas for improvement for security integration with automatic code generation functionality. Advanced risk mitigation is enabled in the framework across multiple process areas, where techniques such as static code analysis, automated penetration testing, and adversarial training hold much promise. The Hybrid ANN-ISM Mechanism is a stable and flexible solution for cybersecurity risk reduction in automatic code generation environments. The coupling of ANN and ISM, in terms of predictive analysis and structured risk management, respectively, contributes effectively towards the security of AI-based code generation tools. More research is required to improve the scalability, privacy preserving, and dynamic integration of the framework with cybersecurity threat intelligence.

Background: Hemodialysis access dysfunction can lead to missed treatments and increased mortality. Traditional monitoring methods, such as physical examination and ultrasound, have limitations, emphasizing the need for a more efficient approach. This study investigates the use of digitized acoustic data to identify and monitor vascular access dysfunction. Methods: This prospective study involved patients undergoing hemodialysis with either arteriovenous fistulas (AVF) or arteriovenous grafts (AVG) between June 2023 and February 2025. All patients underwent vascular imaging (either angiography or ultrasound) to confirm the degree of stenosis. Acoustic data were recorded using a standardized procedure at various puncture sites. Pre- and post-angioplasty data were also collected to assess the effects of vascular intervention. The digitized acoustic data were analyzed for changes in relative loudness, peak-to-valley ratios, and frequency distribution. Results: A total of 157 patients with 236 audio recordings (mean age: 67 ± 11 years; 58% male) were included. Significant acoustic differences were found at the arterial puncture and anastomosis sites in AVF patients with dysfunction, particularly in venous site dysfunction, which exhibited a more pronounced reduction in sound volume and an increased peak-to-valley ratio. After angioplasty, acoustic changes were observed in both arterial and venous sites, with values moving toward normal levels. However, no significant acoustic changes were observed in AVG patients. Additionally, frequency distribution ratios showed minimal clinical relevance. Conclusions: Digitized acoustic data, particularly from the arterial puncture and anastomosis sites, can be an effective tool for detecting and monitoring hemodialysis access dysfunction. These findings suggest potential for acoustic analysis in clinical practice, especially when integrated with AI models for better diagnostics.

This study aims to evaluate the diagnostic accuracy, consistency and diagnostic success rates of eight different AI-based chatbots in Endodontics. This cross-sectional study evaluated diagnostic accuracy of eight diverse AI models, selected for architectural/developer heterogeneity and clinical relevance, using 12 validated fictitious endodontic cases aligned with AAE guidelines and ethical approval was waived as no human data were used. STROBE guidelines were followed to ensure methodological rigor. Standardized prompts ensured uniformity, with three independent executions per case to assess consistency. Responses were anonymized and evaluated by blinded, calibrated reviewers and statistical analysis included Kruskal-Wallis, Dunn’s tests, Fleiss’ Kappa, and chi-square to compare diagnostic/treatment accuracy and intramodel agreement. The analysis revealed significant diagnostic accuracy variation among AI models (p &lt; 0.001), with ChatGPT o1 (97%), Claude (97%), and DeepSeek (90.9%) outperforming Gemini (54.5%). Treatment recommendations showed uniformly high accuracy (97–100%, p = 0.537). Multivariate regression confirmed ChatGPT o1 (OR=32.7) and Claude (OR=30.5) as superior, though complex diagnoses (e.g., acute apical abscess, asymptomatic irreversible pulpitis) reduced accuracy (OR=0.01–0.3, p&lt;0.05). Stratified analysis identified model-specific vulnerabilities: Gemini failed in reversible pulpitis (0/3, p=0.001) and chronic apical abscess (0/3, p=0.001), while ChatGPT o1 struggled with acute apical abscess (0/3, p&lt;0.001). Overall agreement was 93%, with high intraclass reliability (ICC &gt;0.85) for top models versus Gemini (ICC=0.65). Fleiss’ Kappa highlighted moderate agreement (κ=0.28–0.45) in ambiguous cases, emphasizing heterogeneous reliability. In conclusion, seven AI chatbots demonstrated high accuracy in endodontics cases, being considered as helpful tools for complement of clinical practice.

This study reviews the advancements in AI-driven methods for predicting stock prices, tracing their evolution from traditional approaches to modern finance. The role of AI in the market extends beyond predictive systems to encompass the intersection of financial markets with emerging technologies, such as blockchain, and the potential influence of quantum computing on economic modeling. A decentralized finance system examines the application of Reinforcement Learning in financial market prediction, highlighting its potential for continuous learning from dynamic market conditions. The study discusses the development of hybrid prediction models, stock market machine learning systems, and AI-driven investment portfolio management. The potential of quantum computing enhances portfolio analysis, fraud detection, optimization, and asset valuation for complex market predictions, as well as the impact of blockchain technologies on transparency, security, and efficiency. Machine learning techniques can significantly automate data collection and purification. Financial decision-making and the application of time-series analysis techniques can be readily learned through deep reinforcement learning for stock price prediction. Deep Neural Networks and Strategic Asset Allocation can be managed by evaluating performance and portfolio using real-time market insights from AI models. Although there are numerous ethical, sentimental, regulatory, and data quality issues in market prediction, the future job market is heavily dependent on these criteria, particularly through effective risk management and fraud detection.

In a typical engineering design, there are often many design parameters to consider. Also, there are multiple competing requirements and objectives to meet. Manual approach of adjusting the parameters to achieve specific objectives is not optimal especially as the design becomes complex. In the quest for optimizing complex engineering systems, the exploration of the design space becomes imperative, especially when dealing with multi-objective systems characterized by an array of independent variables. This paper presents a comprehensive study on the design space mapping of complex engineering systems, utilizing a turboshaft engine as a case study. The initial phase of our methodology employs a physics-based model to generate synthetic dataset, reflecting the intricate interplay of various system parameters underpinning the engine's operation. This synthesized data serves as a foundation for the subsequent development of a Machine Learning or Deep Learning based surrogate model. The surrogate AI model, will be crafted to encapsulate multiple inputs and outputs inherent in the turboshaft engine's functioning, thereby facilitating an efficient and accurate exploration of the design space. Through this investigation, we will evaluate the efficacy of combining physics-based models with AI-driven techniques in mapping the design space of multi-objective systems. The core of our investigation revolves around the utilization of the AI surrogate model for achieving multi-objective optimization. This optimization process is not only focused on enhancing specific performance metrics but is also geared towards identifying a comprehensive family of feasible design solutions. Such an approach enables the delineation of the entire design space, offering invaluable insights into the trade-offs and synergies among different design objectives. Through this methodology, our goal is to uncover a wide spectrum of viable design alternatives, thereby providing a robust framework for decision-making in the engineering design process.