【202305】美国电子隐私信息中心-生成式人工智能的潜在有害影响与未来之路_86页_英.pdf

CONTRIBUTIONS BY

Grant Fergusson

Caitriona Fitzgerald

Chris Frascella

Megan Iorio

Tom McBrien

Calli Schroeder

Ben Winters

Enid Zhou

Notes on this Paper:

This is version 1 of this paper and is reflective of documented and

anticipated harms of Generative AI as of May 15, 2023. Due to the fast-

changing pace of development, use, and harms of Generative AI, we want to

acknowledge that this is an inherently dynamic paper, subject to changes in

the future.

Throughout this paper, we use a standard format to explain the typology of

harms that generative AI can produce. Each section first explains relevant

background information and potential risks imposed by generative AI, then

highlights specifics harms and interventions that scholars and regulators

and

2. Joy Buolamwini’s Taxonomy of Algorithmic Harms, comprising loss of

opportunity, economic loss, and social stigmatization, including loss of

liberty, increased surveillance, stereotype reinforcement, and other

dignitary harms.

These taxonomies do not necessarily cover all potential AI harms, and our

use of these taxonomies is meant to help readers visualize and

contextualize AI harms without limiting the types and variety of AI harms that

Harassment, Impersonation, and Extortion .............................................................. 9

Spotlight: Section 230 .................................................................................................. 19

Profits Over Privacy: Increased Opaque Data Collection ................................... 24

Increasing Data Security Risk .................................................................................... 30

Confronting Creativity: Impact on Intellectual Property Rights ......................... 33

Exacerbating Effects of Climate Change ................................................................ 40

Labor Manipulation, Theft, and Displacement ....................................................... 44

Spotlight: Discrimination .............................................................................................. 53

The Potential Application of Products Liability Law ............................................. 54

Exacerbating Market Power and Concentration ................................................... 57

OpenAI’s decision to release ChatGPT, a chatbot built on the Large

Language Model GPT-3, last November thrust AI tools to the forefront of

public consciousness. In the last six months, new AI tools used to generate

text, images, video, and audio based on user prompts exploded in

popularity. Suddenly, phrases like Stable Diffusion, Hallucinations, and Value

Alignment were everywhere. Each day, new stories about the different

These

old problems are exacerbated by the industry’s shift in goals from research

and transparency to profit, opacity, and concentration of power. The

widespread availability and hype of these tools has led to increased harm

both individually and on a massive scale. AI replicates racial, gender, and

disability discrimination, and these harms are weaved inextricably through

every issue highlighted in this report.

OpenAI and other companies’ decisions to rapidly integrate generative AI

technology into consumer-facing products and services have undermined

longstanding efforts to make AI development transparent and accountable,

This rapid deployment of generative AI systems without adequate

safeguards is clear evidence that self-regulation has failed. Hundreds of

entities, from corporations to media and government entities, are

developing and looking to rapidly integrate these untested AI tools into a

wide range of systems. And this rapid rollout will have disastrous results

without necessary fairness, accountability, and transparency protections

built in from the beginning.

We are at a critical juncture as policymakers and industry around the globe

are focusing on the substantial risks and opportunities posed by AI. There is

an opportunity to make this technology work for people. Companies should

however, it remains to be seen if their focus will be only on supporting

companies with their development of AI tools and requiring marginal

disclosure and transparency requirements. Enacting clear prohibitions on

high-risk uses, addressing the easy spread of disinformation, requiring

meaningful and proactive disclosures that facilitate informed consent, and

bolstering consumer protection agencies are necessary to address the

harms and risks specific to generative AI. This paper strives to provide a

broad outline of different issues that the use of generative AI brings up,

educate lawmakers and the public, and offer some paths forward to mitigate

harm.