27 lines
2.7 KiB
Markdown
27 lines
2.7 KiB
Markdown
full-title:: [TOWARDS RESPONSIBLE AI IN THE ERA OF GENERATIVE AI: A REFERENCE ARCHITECTURE FOR DESIGNING FOUNDATION MODEL BASED SYSTEMS](https://omnivore.app/me/https-arxiv-org-pdf-2304-11090-18e1add9d8e)
|
|
site:: [arxiv.org](https://arxiv.org/pdf/2304.11090.pdf)
|
|
labels:: [[ResearchPaper]] [[ResponsibleAI]]
|
|
date-saved:: [[07-03-2024]]
|
|
source:: [[Omnivore]]
|
|
state:: [[reading]]
|
|
|
|
- ### Highlights
|
|
collapsed:: true
|
|
- > AI/AGI misuse poses a considerable challenge, which requires continuous risk assessment to ensure the
|
|
>
|
|
instructions for the FM-based systems set by humans are trustworthy and responsible. [⤴️](https://omnivore.app/me/https-arxiv-org-pdf-2304-11090-18e1add9d8e#86951851-e1ae-4ba5-8d70-7a0a5bb0df0a)
|
|
- > foundations models [⤴️](https://omnivore.app/me/https-arxiv-org-pdf-2304-11090-18e1add9d8e#39823a36-af8d-40fc-a7c9-d9f2f46c854a)
|
|
- > architecture evolution of AI systems in the era of foundation models, transitioning
|
|
>
|
|
from ”foundation-model-as-a-connector” to ”foundation-model-as-a-monolithic architecture [⤴️](https://omnivore.app/me/https-arxiv-org-pdf-2304-11090-18e1add9d8e#e807e3f8-96dc-4468-8de2-9d2193cbf418)
|
|
- > pattern-oriented reference architecture for
|
|
>
|
|
designing responsible foundation-model-based syste [⤴️](https://omnivore.app/me/https-arxiv-org-pdf-2304-11090-18e1add9d8e#8cff9ee4-6e1d-43c9-bc5c-2b95321077cb)
|
|
- > adaptability and modifiability. [⤴️](https://omnivore.app/me/https-arxiv-org-pdf-2304-11090-18e1add9d8e#bbce1ca3-b813-43d0-85d6-56cd6c4c28bb)
|
|
- > FMs are massive artificial intelligence (AI) models that are pre-trained on vast
|
|
>
|
|
amounts of broad data and can be adapted to perform a wide variety of tasks [1]. With numerous projects already
|
|
>
|
|
underway to explore their potential, it is widely predicted that FMs will serve as the fundamental building blocks for
|
|
>
|
|
most future AI and artificial generative intelligence (AGI) systems [⤴️](https://omnivore.app/me/https-arxiv-org-pdf-2304-11090-18e1add9d8e#040ebb11-da9d-4ce9-8591-32bdb5112874)
|
|
- > The AI models are responsible for processing data and making inference, while
|
|
>
|
|
the non-AI components are responsible for tasks such as user interface, data storage, interaction with other
|
|
>
|
|
systems. [⤴️](https://omnivore.app/me/https-arxiv-org-pdf-2304-11090-18e1add9d8e#532f4ceb-f80a-49c3-ac69-cae0cad8308a)
|
|
- > enabling accountability necessitates the underlying supporting
|
|
>
|
|
mechanisms for traceability [⤴️](https://omnivore.app/me/https-arxiv-org-pdf-2304-11090-18e1add9d8e#a0300af1-4bf3-42bd-9361-853acdf7fb9d)
|
|
- > trustworthiness is a substantial obstacle in designing the FM-based system [⤴️](https://omnivore.app/me/https-arxiv-org-pdf-2304-11090-18e1add9d8e#f1822cc2-89db-4cb5-bd01-eb13b4b2e132) |