Abstract
Published version: https://scholars.org/contribution/critical-need-transparency-and-regulation As artificial intelligence (AI) technologies cross over a vital threshold of competitiveness with human intelligence, it is necessary to properly frame critical questions in the service of shaping policy and governance while sustaining human values and identity. Given AI's vast socioeconomic implications, government actors and technology creators must proactively address the unique and emerging ethical concerns inherent to AI's many uses. (Samuel, 2021) Open Source versus Black Box AI Technologies AI is an adaptive " set of technologies that mimic the functions and expressions of human intelligence, specifically cognition, and logic " (Samuel, et al., 2022). In the AI field, foundation models (FMs) are more or less what they sound like large, complex models that have been trained on vast quantities of digital general information that may then be adapted for more specific uses (Bommasani, et al., 2021). Two notable features of FMs include a propensity to gain new and often unexpected capabilities as they increase in scale (" emergence ") and a growing predisposition to serve as a common " intelligence base " for differing specialized functions and AI applications (" homogenization ") (Schneider, 2022) Large language models (LLMs) that power applications like ChatGPT are FMs with a focus on modeling human language, knowledge, and logic. Advanced AIs and FMs have the potential to replace multiple task-specific or narrow AIs due to their scale and flexibility, which increases the risk of a few powerful persons or entities who control these advanced AIs gaining extraordinary socioeconomic power, creating conditions for mass exploitation and abuse. ChatGPT and other LLM applications that are growing in popularity use a nontransparent " black box " approach; users of these technologies have little to no access to the inner workings or underlying AI models and can only observe outputs (such as an essay) that result from data inputs (such as a written or spoken prompt) to judge how these applications function. Such opaque FMs have widespread future AI application potential, displaying homogenization where the base models can be adapted to serve a range of specialized purposes. This exponential the risks inherent in a system that allows for private control of advanced AIs. Open-source initiatives, on the other hand, prioritize transparency and public availability of the AI models, including code, process, relevant data, and documentation, so that users and society at large have an opportunity to understand how these technologies function. For human society, the spirit of the open-source movement is one of the most valuable forces in the AI technologies arena (The open source definition,2023). Research and development of AI ethics must emphasize the contributions of open data, open-source software, open knowledge, and responsible AI movements and contrast these with the challenges presented by relatively opaque AI applications (Weber, 2009; Franco-Bedoya, et al., 2017). Closely coupled with open-source, the open-data movement (which refers to " data that is made freely available for open consumption, at no direct cost to the public, which can be efficiently located, filtered, downloaded, processed, shared, and reused without any significant restrictions on associated derivatives, use, and reuse ") can be a significant contributor to the development of responsible AI (Chidipothu, et al., 2022). Open-source initiatives distribute power and reduce the likelihood of centralized control and abuse by a few AI owners with concentrated power.