Christopher Bogart

Jaromir Savelka, Arav Agarwal, Marshall An et al. · cmu

This paper studies recent developments in large language models' (LLM) abilities to pass assessments in introductory and intermediate Python programming courses at the postsecondary level. The emergence of ChatGPT resulted in heated debates of its potential uses (e.g., exercise generation, code explanation) as well as misuses in programming classes (e.g., cheating). Recent studies show that while the technology performs surprisingly well on diverse sets of assessment instruments employed in typical programming classes the performance is usually not sufficient to pass the courses. The release of GPT-4 largely emphasized notable improvements in the capabilities related to handling assessments originally designed for human test-takers. This study is the necessary analysis in the context of this ongoing transition towards mature generative AI systems. Specifically, we report the performance of GPT-4, comparing it to the previous generations of GPT models, on three Python courses with assessments ranging from simple multiple-choice questions (no code involved) to complex programming projects with code bases distributed into multiple files (599 exercises overall). Additionally, we analyze the assessments that were not handled well by GPT-4 to understand the current limitations of the model, as well as its capabilities to leverage feedback provided by an auto-grader. We found that the GPT models evolved from completely failing the typical programming class' assessments (the original GPT-3) to confidently passing the courses with no human involvement (GPT-4). While we identified certain limitations in GPT-4's handling of MCQs and coding exercises, the rate of improvement across the recent generations of GPT models strongly suggests their potential to handle almost any type of assessment widely used in higher education programming courses. These findings could be leveraged by educators and institutions to adapt the design of programming assessments as well as to fuel the necessary discussions into how programming classes should be updated to reflect the recent technological developments. This study provides evidence that programming instructors need to prepare for a world in which there is an easy-to-use widely accessible technology that can be utilized by learners to collect passing scores, with no effort whatsoever, on what today counts as viable programming knowledge and skills assessments.

Jaromir Savelka, Arav Agarwal, Christopher Bogart et al. · cmu

We evaluated the capability of generative pre-trained transformers (GPT), to pass assessments in introductory and intermediate Python programming courses at the postsecondary level. Discussions of potential uses (e.g., exercise generation, code explanation) and misuses (e.g., cheating) of this emerging technology in programming education have intensified, but to date there has not been a rigorous analysis of the models' capabilities in the realistic context of a full-fledged programming course with diverse set of assessment instruments. We evaluated GPT on three Python courses that employ assessments ranging from simple multiple-choice questions (no code involved) to complex programming projects with code bases distributed into multiple files (599 exercises overall). Further, we studied if and how successfully GPT models leverage feedback provided by an auto-grader. We found that the current models are not capable of passing the full spectrum of assessments typically involved in a Python programming course (<70% on even entry-level modules). Yet, it is clear that a straightforward application of these easily accessible models could enable a learner to obtain a non-trivial portion of the overall available score (>55%) in introductory and intermediate courses alike. While the models exhibit remarkable capabilities, including correcting solutions based on auto-grader's feedback, some limitations exist (e.g., poor handling of exercises requiring complex chains of reasoning steps). These findings can be leveraged by instructors wishing to adapt their assessments so that GPT becomes a valuable assistant for a learner as opposed to an end-to-end solution.

Pragnya Sridhar, Aidan Doyle, Arav Agarwal et al. · cmu

We evaluated the capability of a generative pre-trained transformer (GPT-4) to automatically generate high-quality learning objectives (LOs) in the context of a practically oriented university course on Artificial Intelligence. Discussions of opportunities (e.g., content generation, explanation) and risks (e.g., cheating) of this emerging technology in education have intensified, but to date there has not been a study of the models' capabilities in supporting the course design and authoring of LOs. LOs articulate the knowledge and skills learners are intended to acquire by engaging with a course. To be effective, LOs must focus on what students are intended to achieve, focus on specific cognitive processes, and be measurable. Thus, authoring high-quality LOs is a challenging and time consuming (i.e., expensive) effort. We evaluated 127 LOs that were automatically generated based on a carefully crafted prompt (detailed guidelines on high-quality LOs authoring) submitted to GPT-4 for conceptual modules and projects of an AI Practitioner course. We analyzed the generated LOs if they follow certain best practices such as beginning with action verbs from Bloom's taxonomy in regards to the level of sophistication intended. Our analysis showed that the generated LOs are sensible, properly expressed (e.g., starting with an action verb), and that they largely operate at the appropriate level of Bloom's taxonomy, respecting the different nature of the conceptual modules (lower levels) and projects (higher levels). Our results can be leveraged by instructors and curricular designers wishing to take advantage of the state-of-the-art generative models to support their curricular and course design efforts.

Jaromir Savelka, Arav Agarwal, Christopher Bogart et al. · cmu

We analyzed effectiveness of three generative pre-trained transformer (GPT) models in answering multiple-choice question (MCQ) assessments, often involving short snippets of code, from introductory and intermediate programming courses at the postsecondary level. This emerging technology stirs countless discussions of its potential uses (e.g., exercise generation, code explanation) as well as misuses in programming education (e.g., cheating). However, the capabilities of GPT models and their limitations to reason about and/or analyze code in educational settings have been under-explored. We evaluated several OpenAI's GPT models on formative and summative MCQ assessments from three Python courses (530 questions). We found that MCQs containing code snippets are not answered as successfully as those that only contain natural language. While questions requiring to fill-in a blank in the code or completing a natural language statement about the snippet are handled rather successfully, MCQs that require analysis and/or reasoning about the code (e.g., what is true/false about the snippet, or what is its output) appear to be the most challenging. These findings can be leveraged by educators to adapt their instructional practices and assessments in programming courses, so that GPT becomes a valuable assistant for a learner as opposed to a source of confusion and/or potential hindrance in the learning process.

Christopher Bogart, Aparna Warrier, Arav Agarwal et al.

As artificial intelligence (AI) systems become ubiquitous in professional contexts, there is an urgent need to equip workers, often with backgrounds outside of STEM, with the skills to use these tools effectively as well as responsibly, that is, to be AI literate. However, prevailing definitions and therefore assessments of AI literacy often emphasize foundational technical knowledge, such as programming, mathematics, and statistics, over practical knowledge such as interpreting model outputs, selecting tools, or identifying ethical concerns. This leaves a noticeable gap in assessing someone's AI literacy for real-world job use. We propose a work-task-oriented assessment model for AI literacy which is grounded in the competencies required for effective use of AI tools in professional settings. We describe the development of a novel AI literacy assessment instrument, and accompanying formative assessments, in the context of a US Navy robotics training program. The program included training in robotics and AI literacy, as well as a competition with practical tasks and a multiple choice scenario task meant to simulate use of AI in a job setting. We found that, as a measure of applied AI literacy, the competition's scenario task outperformed the tests we adopted from past research or developed ourselves. We argue that when training people for AI-related work, educators should consider evaluating them with instruments that emphasize highly contextualized practical skills rather than abstract technical knowledge, especially when preparing workers without technical backgrounds for AI-integrated roles.

David Reid, Kalvin Eng, Chris Bogart et al.

This paper presents results from the MSR 2021 Hackathon. Our team investigates files/projects that contain known security vulnerabilities and how widespread they are throughout repositories in open source software. These security vulnerabilities can potentially be propagated through code reuse even when the vulnerability is fixed in different versions of the code. We utilize the World of Code infrastructure to discover file-level duplication of code from a nearly complete collection of open source software. This paper describes a method and set of tools to find all open source projects that use known vulnerable files and any previous revisions of those files.

Yuxing Ma, Tapajit Dey, Chris Bogart et al.

Open source software (OSS) is essential for modern society and, while substantial research has been done on individual (typically central) projects, only a limited understanding of the periphery of the entire OSS ecosystem exists. For example, how are the tens of millions of projects in the periphery interconnected through. technical dependencies, code sharing, or knowledge flow? To answer such questions we: a) create a very large and frequently updated collection of version control data in the entire FLOSS ecosystems named World of Code (WoC), that can completely cross-reference authors, projects, commits, blobs, dependencies, and history of the FLOSS ecosystems and b) provide capabilities to efficiently correct, augment, query, and analyze that data. Our current WoC implementation is capable of being updated on a monthly basis and contains over 18B Git objects. To evaluate its research potential and to create vignettes for its usage, we employ WoC in conducting several research tasks. In particular, we find that it is capable of supporting trend evaluation, ecosystem measurement, and the determination of package usage. We expect WoC to spur investigation into global properties of OSS development leading to increased resiliency of the entire OSS ecosystem. Our infrastructure facilitates the discovery of key technical dependencies, code flow, and social networks that provide the basis to determine the structure and evolution of the relationships that drive FLOSS activities and innovation.

Sadika Amreen, Audris Mockus, Chris Bogart et al.

Graphs of developer networks are important for software engineering research and practice. For these graphs to realistically represent the networks, accurate developer identities are imperative. We aim to identify developer identity errors from open source software repositories in VCS, investigate the nature of these errors, design corrective algorithms, and estimate the impact of the errors on networks inferred from this data. We investigate these questions using over 1B Git commits with over 23M recorded author identities. By inspecting the author strings that occur most frequently, we group identity errors into categories. We then augment the author strings with 3 behavioral fingerprints: time-zone frequencies, the set of files modified, and a vector embedding of the commit messages. We create a manually validated set of identities for a subset of OpenStack developers using an active learning approach and use it to fit supervised learning models to predict the identities for the remaining author strings in OpenStack. We compare these predictions with a commercial effort and a leading research method. Finally, we compare network measures for file-induced author networks based on corrected and raw data. We find commits done from different environments, misspellings, organizational IDs, default values, and anonymous IDs to be the major sources of errors. We also find supervised learning methods to reduce errors by several times in comparison to existing methods and the active learning approach to be an effective way to create validated datasets and that correction of developer identity has a large impact on the inference of the social network. We believe that our proposed Active Learning Fingerprint Based Anti-Aliasing (ALFAA) approach will expedite research progress in the software engineering domain for applications that depend upon graphs of developers or other social networks.

Jacob Doughty, Zipiao Wan, Anishka Bompelli et al. · cmu

There is a constant need for educators to develop and maintain effective up-to-date assessments. While there is a growing body of research in computing education on utilizing large language models (LLMs) in generation and engagement with coding exercises, the use of LLMs for generating programming MCQs has not been extensively explored. We analyzed the capability of GPT-4 to produce multiple-choice questions (MCQs) aligned with specific learning objectives (LOs) from Python programming classes in higher education. Specifically, we developed an LLM-powered (GPT-4) system for generation of MCQs from high-level course context and module-level LOs. We evaluated 651 LLM-generated and 449 human-crafted MCQs aligned to 246 LOs from 6 Python courses. We found that GPT-4 was capable of producing MCQs with clear language, a single correct choice, and high-quality distractors. We also observed that the generated MCQs appeared to be well-aligned with the LOs. Our findings can be leveraged by educators wishing to take advantage of the state-of-the-art generative models to support MCQ authoring efforts.

Atharva Naik, Jessica Ruhan Yin, Anusha Kamath et al. · cmu

An advantage of Large Language Models (LLMs) is their contextualization capability - providing different responses based on student inputs like solution strategy or prior discussion, to potentially better engage students than standard feedback. We present a design and evaluation of a proof-of-concept LLM application to offer students dynamic and contextualized feedback. Specifically, we augment an Online Programming Exercise bot for a college-level Cloud Computing course with ChatGPT, which offers students contextualized reflection triggers during a collaborative query optimization task in database design. We demonstrate that LLMs can be used to generate highly situated reflection triggers that incorporate details of the collaborative discussion happening in context. We discuss in depth the exploration of the design space of the triggers and their correspondence with the learning objectives as well as the impact on student learning in a pilot study with 34 students.

Jaromir Savelka, Can Kultur, Arav Agarwal et al. · cmu

The accelerating pace of developments in Artificial Intelligence~(AI) and the increasing role that technology plays in society necessitates substantial changes in the structure of the workforce. Besides scientists and engineers, there is a need for a very large workforce of competent AI technicians (i.e., maintainers, integrators) and users~(i.e., operators). As traditional 4-year and 2-year degree-based education cannot fill this quickly opening gap, alternative training methods have to be developed. We present the results of the first four years of the AI Technicians program which is a unique collaboration between the U.S. Army's Artificial Intelligence Integration Center (AI2C) and Carnegie Mellon University to design, implement and evaluate novel rapid occupational training methods to create a competitive AI workforce at the technicians level. Through this multi-year effort we have already trained 59 AI Technicians. A key observation is that ongoing frequent updates to the training are necessary as the adoption of AI in the U.S. Army and within the society at large is evolving rapidly. A tight collaboration among the stakeholders from the army and the university is essential for successful development and maintenance of the training for the evolving role. Our findings can be leveraged by large organizations that face the challenge of developing a competent AI workforce as well as educators and researchers engaged in solving the challenge.

Margaret Burnett, Martin Erwig, Abrar Fallatah et al.

HCI researchers' and practitioners' awareness of intersectionality has been expanding, producing knowledge, recommendations, and prototypes for supporting intersectional populations. However, doing intersectional HCI work is uniquely expensive: it leads to a combinatorial explosion of empirical work (expense 1), and little of the work on one intersectional population can be leveraged to serve another (expense 2). In this paper, we explain how representations employed by certain analytical design methods correspond to type abstractions, and use that correspondence to identify a (de)compositional model in which a population's diverse identity properties can be joined and split. We formally prove the model's correctness, and show how it enables HCI designers to harness existing analytical HCI methods for use on new intersectional populations of interest. We illustrate through four design use-cases, how the model can reduce the amount of expense 1 and enable designers to leverage prior work to new intersectional populations, addressing expense 2.

Sreecharan Sankaranarayanan, Siddharth Reddy Kandimalla, Christopher Bogart et al.

Contributing to the literature on aptitude-treatment interactions between worked examples and problem-solving, this paper addresses differential learning from the two approaches when students are positioned as domain experts learning new concepts. Our evaluation is situated in a team project that is part of an advanced software engineering course. In this course, students who possess foundational domain knowledge but are learning new concepts engage alternatively in programming followed by worked example-based reflection. They are either allowed to finish programming or are curtailed after a pre-specified time to participate in a longer worked example-based reflection. We find significant pre- to post-test learning gains in both conditions. Then, we not only find significantly more learning when students participated in longer worked example-based reflections but also a significant performance improvement on a problem-solving transfer task. These findings suggest that domain experts learning new concepts benefit more from worked example-based reflections than from problem-solving.

Christopher Bogart, James Howison, James Herbsleb

Software is increasingly produced in the form of ecosystems, collections of interdependent components maintained by a distributed community. These ecosystems act as network organizations, not markets, and thus often lack actionable price-like signals about how the software is used and what impact it has. We introduce a tool, the Scientific Software Network Map, that collects and displays summarized usage data tailored to the needs of actors in software ecosystems. We performed a contextualized walkthrough of the Map with producers and stewards in six scientific software ecosystems that use the R language. We found that they work to maximize diversity rather than quantity of uses, and to minimize coordination costs. We also found that summarized usage data would be useful for justifying ecosystem work to funding agencies; and we discovered a variety of more granular usage needs that would help in adding or maintaining features.

Zack Coker, David Gray Widder, Claire Le Goues et al.

Aspects of frameworks, such as inversion of control and the structure of framework applications, require developers to adjust their debugging strategies as compared to debugging sequential programs. However, the benefits and challenges of framework debugging are not fully understood, and gaining this knowledge could provide guidance in debugging strategies and framework tool design. To gain insight into the process developers use to fix problems in framework applications, we performed two human studies investigating how developers fix applications that use a framework API incorrectly. These studies focused on the Android Fragment class and the ROS framework. We analyzed the results of the studies using a mixed-methods approach, consisting of techniques from grounded theory, qualitative content analysis, and case studies. From our analysis, we produced a theory of the benefits and challenges of framework debugging. This theory states that developers find inversion of control challenging when debugging but find the structure of framework applications helpful. This theory could be used to guide strategies for debugging framework applications and framework tool designs.

Shrimai Prabhumoye, Samridhi Choudhary, Evangelia Spiliopoulou et al.

There has been a long standing interest in understanding `Social Influence' both in Social Sciences and in Computational Linguistics. In this paper, we present a novel approach to study and measure interpersonal influence in daily interactions. Motivated by the basic principles of influence, we attempt to identify indicative linguistic features of the posts in an online knitting community. We present the scheme used to operationalize and label the posts with indicator features. Experiments with the identified features show an improvement in the classification accuracy of influence by 3.15%. Our results illustrate the important correlation between the characteristics of the language and its potential to influence others.