62588 research outputs found
Sort by
(Un)Ethical AI: Fact and Fiction - Project Summary
No full textThis lesson template provides a non-technical introduction to ethical considerations about AI, through the lens of a Captain America film.We are bombarded with claims about the ways that AI algorithms are remaking society, often made with hyperbolic comparisons to science fiction. How should we evaluate these claims? Though they hear about the impact of "algorithms" all the time, students often do not know what they are or how they work, unless they are pursuing courses in computer science. In order to effectively evaluate the news around them, and understand which tools are valuable and which are troubling, students in nontechnical fields need to understand what an algorithm is and how one can be ethically deployed.
This lesson is designed to introduce the concept of ethical AI through the movie Captain America: Winter Soldier, which has a subplot about an AI-powered assassination device, and compare this fictional tool to real-world cases including social network recommender algorithms, FICO credit scores, and automated resume classification programs. The lesson will include a conceptual description of a neural network, with an emphasis on training data and the limitations of systems that aim to predict the future based on data from the past. Knowledge of coding or mathematical principles will not be assumed. Instructors can include additional or alternative examples of AI in pop culture at their discretion.
Students will be able to describe how a neural network works, including the concepts of training data, classification, and measurement error. They will also be able to discuss the ethical principles behind gathering data, training models, and deploying them. They will be able to evaluate for themselves whether a model was ethically developed and whether, and in what contexts, it should be deployed
Future-Proofing AI at Lehigh University: A Guide to LLM Evaluation and Usage - Project Summary
No full textDevelop a LLM model evaluation guide, framework, and workshop to promote AI and prompt engineering literacy at Lehigh University.The rapidly evolving AI landscape presents continuous advancements in performance, features, and functionality. Vendors frequently release new updates, often outpacing each other within a
span of months. For instance, even within a single vendor like OpenAI, the release cadence is challenging to follow.
Although vendors provide performance charts for their models, these metrics may not accurately reflect real-world usage. This creates a challenge for faculty, researchers, and staff, such as Library & Technology Services, who must keep abreast of these changes to identify the best model for specific projects or to re-evaluate existing projects.
To address this challenge, it is crucial to establish a systematic approach for tracking and evaluating AI model updates and their practical applications. This will enable informed decision-making and ensure the optimal selection of AI models for various academic, research, and staff needs.
The proposed solution is to develop a LLM evaluation framework that can be used
against various models, document new features, keep track of performance
and the quality of the responses. Look to develop a tool that can capture test
prompts, run those prompts, collect analysis data, define quality metrics, and
store for historical references. Provide support for both human and LLM graders.
The delivery includes the following:
1. Process framework to help provide guidelines in evaluating and selecting LLMs to be used for academics, research, or general staff usage.
2. Tool that can be used to develop prompts for testing and evaluation to determine performance and response quality, including hallucinations.
3. Provide documentation on the framework, tool, and offer prompt engineering and evaluation workshops as needed.
The goal is to assist the Lehigh community in evaluating LLMs in this rapidly changing environment to make informed and optimal selection for their projects or usage. In addition, promote a deeper understanding of prompt engineering, a crucial skill for effectively interacting with AI models
DoppelgAIngers
No full textDoppelgAIngers displays a false "mirror" image that looks eerily like you: dressed similarly, posed similarly, but who does not exist."Deep fakes" are images or videos that have been manipulated with AI to misrepresent the truth. These images spread virally online, giving credence to conspiracy theories and stoking political, social, and cultural tensions. These images force us to question the nature of visual truth. How can we investigate what we know to be true? How can we reconcile the reality of the physical world with the subjective nature of images? How can we recognize manipulated images when we see them? DoppelgAIngers is a web-based interactive piece and series of images. This work raises questions about truth in imagery and highlights the uncanny nature of AI-generated images. The piece functions like a "camera" or "mirror" but instead of reflecting a true image, it displays an image that has been passed through AI as if in a game of telephone. The image or video input is described by AI and that image description is fed back into AI as an image prompt. For instance, a person stands in front of a web camera to be met with an unsettling AI doppleganger — a person who looks eerily like them, wearing a similar outfit, posed in a similar way, but who does not exist. By connecting AI images to personal identity and understanding of self, the work intends to raise conversation around ethics and media literacy. It can be used as a tool for conversation, community engagement, and education around AI images.</p
Entrepreneurial Edge: Harnessing AI to Revolutionize Sales with Creative Intelligence - Project Summary
No full textStudents engage in a multi-week sales project, using Generative AI to develop sales strategies while considering ethical implications.Challenge (or Opportunity) // Many universities don\u27t specifically teach the sales process. However, creating and applying innovative sales strategies can enhance an emerging entrepreneur\u27s confidence, reach more segments of the target market, and speed up organic growth of the company. Sales skills are often best acquired through hands-on experience, involving trial and error, reflection, and adaptation.
Proposed solution // Through a multi-week, course-based assignment, students are engaging in a creative-approach to the sales process. Inspired by the \u27one red paperclip\u27 project, each student was provided one paperclip during the first week of class. They spent the past week \u27trading up\u27 for progressively bigger and better items, articulating their lessons learned along the way. Moving forward is where it gets really interesting! They will continue to trade up over the next few weeks, utilizing Generative AI as their sales coach, creative partner, and tool to boost trades (aka sales) to levels they never knew were possible! A discussion of ethical use of AI as a sales tool will be integrated.
Delivery
1. Get the trading (sales) process started without the aid of AI. Reflect on what worked for the sequential trades (sales), as well as where and what obstacles surfaced.
2. Add AI to the mix of the trading (sales) process, encouraging students to use the technology to help write compelling product descriptions, identify aspirational \u27one-level-up\u27 trades, and develop previously unconsidered trading strategies.
3. Each week students will share how AI helped (or hindered) their trading (sales) process.
4. Integrate a classroom roundtable discussion on any unanticipated ethical dilemmas during the assignment, focusing on the ethical use of AI as a sales tool.
Outcomes // Better skills in sales with our emerging entrepreneurs
An Investigation of The Effect of Rheo-Printing Technology on Big Area Additive Manufacturing
No full textA novel processing innovation named Rheo-printing technology was introduced that could impact the field of the Big Area Additive Manufacturing (BAAM). The Rheo-printing technology applies a controlled circumferential and axial shear rate to the polymer melt before depositing the polymer through the printing nozzle. The rheological polymer properties modify due to the applied shear rate, and the ultimate goal is to enhance the product\u27s properties. The application of the circumferential shear rate on the polymer melt is accomplished through a rotational printing nozzle. Adjusting the rotational speed of the printing nozzle controls the shear rate applied to the polymer melt, providing control over the rheology of the polymer melt. This research employs an extrusion-based AM machine that uses polymer pellets as a feedstock to investigate the impact of Rheo-printing technology on BAAM.The effect of Rheo-printing technology was investigated theoretically and numerically to examine the shear rate impact on the material\u27s viscosity. Different shear-thinning polymers were included in the investigation; the viscosity influence of each polymer has been studied through a range of rotational speeds, and statistical analysis has been applied to the obtained results for a comprehensive understanding. The investigation compared the influence of Rheo-printing technology on two different nozzle sizes. For small additive manufacturing or desktop 3D printers, a 0.6 mm nozzle diameter was utilized, while a 2 mm nozzle diameter was employed for BAAM. Overall results showed that the effect of Rheo-printing technology appeared more significant with a bigger nozzle diameter. Also, there is a favorable correlation between nozzle rotation and viscosity for all polymers, regardless of nozzle diameter. Specifically, this correlation manifests as a decrease in viscosity as the nozzle is rotated, with the magnitude of this reduction becoming more pronounced at higher rotational speeds and near in the outermost region of the extruded polymer road. Two different numerical simulations were included to study the impact of the temperature on the printing process. The first numerical simulation was to study the effect of printing speed on the temperature evolution of each layer during the printing process. Results showed that the temperature fluctuation as each layer\u27s heating and cooling during the printing process decreased with increasing the printing speed. The second numerical simulation was to study the platform temperature\u27s impact on each layer\u27s temperature evolution during printing. The effect of increasing the platform temperature was shown clearly on the cooling interval of each layer. Increasing platform temperature reduces the heat losses in the printed layer during the cooling interval. An experimental investigation also employed conventional printing and Rheo-printing technology to investigate the temperature evolution during printing and compare the obtained results. This investigation was performed by printing two samples using conventional and Rheo-printing technology and monitoring the temperature evolution during printing. The anisotropy and porosity of BAAM products were investigated, and the enhancement of Rheo-printing technology on product properties was validated numerically and experimentally. Three groups of different layer-building times were designed, and twenty-four samples were printed using conventional and Rheo-printing technology, investigating the impact of Rheo-printing technology on interlayer adhesion strength. Also, three configurations were designed, and eighteen samples were printed using conventional and Rheo-printing technology, investigating the effect of Rheo-printing technology on void formation. The enhancement of the Rheo-printing technology on mechanical properties of BAAM samples was validated where the anisotropy and porosity percentage were reduced
Advances in Discrete Optimization: from Truss Design Optimization to Quantum Computing Optimization
No full textThis manuscript explores optimization problems and spans two distinct but interconnected parts. The first part delves into the practical application of mathematical optimization techniques with a particular focus on discrete structural design optimization problems, notorious for their combinatorial, nonlinear, and non-convex nature. The second part focuses on quantum computing, examining its potential for solving optimization problems.Chapter 3 presents two mathematical formulations for structural design optimization. These formulations are designed to handle discrete cross-sectional areas. The chapter proposes rigorous mathematical approaches to address the stability of structural configurations. We leverage the linear and bilinear nature of these problems to exploit the rescaling properties of both the design and auxiliary variables while also extending the superposition principle to accommodate nonlinear stress constraints.In Chapter 4, we introduce the neighborhood search mixed-integer linear optimization (NS-MILO) method. This method is developed based on the insights gained from the preceding chapter, leveraging the specific characteristics of the optimization problems discussed. This chapter comprises a comprehensive set of experiments designed to provide compelling empirical evidence regarding the effectiveness of the proposed solution methodologies.In Chapter 5, our focus shifts to a challenging problem known as the max k-cut problem, a problem of considerable complexity within combinatorial optimization. Within this chapter, we undertake a systematic examination of various optimization formulations tailored to address this problem while rigorously assessing their practical efficacy.Additionally, the chapter extends its exploration beyond traditional optimization formulations, delving into the domain of binary quadratic optimization () formulations and quantum-inspired methodologies. These novel approaches represent an intriguing avenue for addressing the max k-cut problem, harnessing insights from quantum computing principles without making any claims of surpassing classical methods.Finally, Chapter~
ef{chapter: Quantum Linear Optimization} delves into inexact interior-point methods for linear optimization problems, considering the potential integration of quantum linear system solvers. It highlights the advantages and challenges posed by quantum solvers and investigates iterative refinement techniques to enhance their performance
Exploring the Impact of Sentiment Analysis on Price Prediction
No full textThe integration of sentiment analysis into predictive models for financial markets, particularly Bitcoin, combines behavioral finance with quantitative analysis. This thesis investigates the extent to which sentiment data, derived from social media platforms such as X (formerly Twitter), can enhance the accuracy of Bitcoin price predictions. A key idea in the study is that public sentiment, as shown on social media, affects Bitcoin\u27s market prices. The research uses linear regression models that combine Bitcoin\u27s opening prices with sentiment scores from social media to forecast closing prices. The analysis covers the period from January 2012 to December 2019. Sentiment scores were analyzed using VADER and TextBlob lexicons. The empirical findings show that models incorporating sentiment scores enhance predictive accuracy. For example, incorporating daily average sentiment scores v_avg and B_avg into the models reduced the Mean Squared Error (MSE) from 81184 to 81129 and improved other metrics such as Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE), particularly at specific lag times like 8 and 76 days. These results emphasize the potential benefits of sentiment analysis to improve financial forecasting models. However, it also acknowledges limitations related to the scope of data and the complexities of accurately measuring sentiment. Future research is encouraged to explore more sophisticated models and diverse data sources to further enhance and validate the integration of sentiment analysis in financial forecasting
