AI as an ecosystem: ensuring generative AI Is safe and effective

Enrico Coiera; David Fraile Navarro

AI as an ecosystem: ensuring generative AI Is safe and effective

Enrico Coiera, David Fraile Navarro

Australian Institute of Health Innovation

Research output: Contribution to journal › Article

173 Downloads (Pure)

Abstract

Traditional scientific methods may not be sufficient to evaluate generative artificial intelligence (GenAI) due to its nondeterministic performance and lack of structural transparency. Viewing GenAI as an ecosystem may facilitate global performance analysis, such as estimating resilience and sustainability as conditions or tasks change. Adopting GenAI necessitates a shift in evaluation focus from reliance on premarket assessment to real- world postmarket surveillance.

Original language	English
Journal	NEJM AI
Publication status	Accepted/In press - 8 Jul 2024

Access to Document

Accepted Author version (open access)Accepted author manuscript, 465 KB

Cite this

@article{37f97bdcc2fe4c418ee37f05ccf6dab0,

title = "AI as an ecosystem: ensuring generative AI Is safe and effective",

abstract = "Traditional scientific methods may not be sufficient to evaluate generative artificial intelligence (GenAI) due to its nondeterministic performance and lack of structural transparency. Viewing GenAI as an ecosystem may facilitate global performance analysis, such as estimating resilience and sustainability as conditions or tasks change. Adopting GenAI necessitates a shift in evaluation focus from reliance on premarket assessment to real- world postmarket surveillance.",

author = "Enrico Coiera and {Fraile Navarro}, David",

year = "2024",

month = jul,

day = "8",

language = "English",

journal = "NEJM AI",

issn = "2836-9386",

publisher = "Massachussetts Medical Society",

}

TY - JOUR

T1 - AI as an ecosystem

T2 - ensuring generative AI Is safe and effective

AU - Coiera, Enrico

AU - Fraile Navarro, David

PY - 2024/7/8

Y1 - 2024/7/8

N2 - Traditional scientific methods may not be sufficient to evaluate generative artificial intelligence (GenAI) due to its nondeterministic performance and lack of structural transparency. Viewing GenAI as an ecosystem may facilitate global performance analysis, such as estimating resilience and sustainability as conditions or tasks change. Adopting GenAI necessitates a shift in evaluation focus from reliance on premarket assessment to real- world postmarket surveillance.

AB - Traditional scientific methods may not be sufficient to evaluate generative artificial intelligence (GenAI) due to its nondeterministic performance and lack of structural transparency. Viewing GenAI as an ecosystem may facilitate global performance analysis, such as estimating resilience and sustainability as conditions or tasks change. Adopting GenAI necessitates a shift in evaluation focus from reliance on premarket assessment to real- world postmarket surveillance.

M3 - Article

SN - 2836-9386

JO - NEJM AI

JF - NEJM AI

ER -