notesum.ai

Published at December 10

"So what if I used GenAI?" -- Implications of Using Cloud-based GenAI in Software Engineering Research

cs.SE

Released Date: December 10, 2024

Authors: Gouri Ginde¹

Aff.: ¹University of Calgary

Arxiv: http://arxiv.org/pdf/2412.07221v1

Refer to caption

	Open-Source	Free-Tier (cloud-based)	Enterprise (license-based)
Accessibility	Publicly available with open licensing	Limited free access via APIs/interfaces	Subscription-based access
Customization	Fully customizable; fine-tuning supported.	Limited to prompt engineering; no fine-tuning.	Domain-specific tuning.
Data Privacy	Full control; user responsibility for security	Data may be retained for model improvements	Guaranteed privacy
Cost	Free to use; hosting and training incur costs	Free but with usage limits	Subscription pricing
Ease of Use	Requires ML expertise to deploy and manage	Beginner-friendly; no setup needed	Seamless integration options.
Performance	Varies by model; requires optimization.	Optimized for general-purpose tasks.	High performance, tailored for business needs.
Examples	LLaMA, Falcon, GPT-J, GPT-NeoX, Mistral.	ChatGPT Free, Google Bard, Claude Instant.	ChatGPT Enterprise, Azure OpenAI