Spaces:
Running
Running
File size: 14,225 Bytes
ac25a9d 80bef93 ac25a9d fd6efb2 80bef93 ac25a9d 80bef93 ac25a9d b1a18ac ac25a9d b55b835 fd6efb2 96c888b b55b835 96c888b ac25a9d fd6efb2 ac25a9d fd6efb2 ac25a9d 96c888b ac25a9d 96c888b ac25a9d fd6efb2 ac25a9d fd6efb2 96c888b fd6efb2 96c888b fd6efb2 37d96b6 fd6efb2 80bef93 96c888b 80bef93 fd6efb2 37d96b6 fd6efb2 96c888b fd6efb2 37d96b6 fd6efb2 404320b ac25a9d 96c888b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 |
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="description"
content="Evaluating Evaluations: Examining Best Practices for Measuring Broader Impacts of Generative AI">
<meta name="keywords" content="Generative AI, Evaluation, Social Impact, NeurIPS, Workshop, AI Ethics">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Evaluating Evaluations: NeurIPS Workshop 2024</title>
<link rel="preconnect" href="https://fonts.googleapis.com">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link href="https://fonts.googleapis.com/css2?family=Libre+Franklin:wght@400;600&display=swap" rel="stylesheet">
<link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro"
rel="stylesheet">
<link rel="stylesheet" href="./static/css/bulma.min.css">
<link rel="stylesheet" href="./static/css/bulma-carousel.min.css">
<link rel="stylesheet" href="./static/css/bulma-slider.min.css">
<link rel="stylesheet" href="./static/css/fontawesome.all.min.css">
<link rel="stylesheet"
href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css">
<link rel="stylesheet" href="./static/css/index.css">
<link rel="icon" href="./static/images/favicon.svg">
<script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
<script defer src="./static/js/fontawesome.all.min.js"></script>
<script src="./static/js/bulma-carousel.min.js"></script>
<script src="./static/js/bulma-slider.min.js"></script>
<script src="./static/js/index.js"></script>
</head>
<body>
<section class="hero">
<div class="hero-body">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-1 publication-title">Evaluating Evaluations</h1>
<h2 class="subtitle is-3 publication-subtitle">Examining Best Practices for Measuring Broader Impacts of Generative AI</h2>
<div class="is-size-5 publication-authors">
<span class="author-block">Workshop co-located with <a href="https://neurips.cc/" target="_blank">NeurIPS 2024</a></span>
</div>
<div class="is-size-5 publication-authors">
<span class="author-block">[Date + maybe zoom link]</span>
</div>
</div>
</div>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Workshop Overview</h2>
<div class="content has-text-justified">
<p>
Generative AI systems are becoming increasingly prevalent in society, producing content such as text, images, audio, and video with far-reaching implications. While the NeurIPS Broader Impact statement has notably shifted norms for AI publications to consider negative societal impact, no standard exists for how to approach these impact assessments. This workshop aims to address this critical gap by bringing together experts on evaluation science and practitioners who develop and analyze technical systems.
</p>
<p>
Building upon our previous initiatives, including the FAccT 2023 CRAFT session "Assessing the Impacts of Generative AI Systems Across Modalities and Society" and our initial "Evaluating the Social Impact of Generative AI Systems" report, we have made significant strides in this area. Through these efforts, we collaboratively developed an evaluation framework and guidance for assessing generative systems across modalities. We have since crowdsourced evaluations and analyzed gaps in the literature and systemic issues around how evaluations are designed and selected, resulting in a more comprehensive second edition of the paper.
</p>
<p>
The goal of this workshop is to share existing findings with the NeurIPS community and collectively develop future directions for effective community-driven evaluations. A key focus is participatory AI: Wide benefits can be gained from the scope and involving all participants, not just domain experts. By encouraging collaboration among experts, practitioners, and the wider community, the workshop aims to create more comprehensive evaluations and develop urgent policy recommendations for governments and AI safety organizations.
</p>
</div>
</div>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<h2 class="title is-3">Workshop Objectives</h2>
<div class="content">
<ol>
<li>Share existing findings and methodologies with the NeurIPS community</li>
<li>Collectively develop future directions for effective community-built evaluations</li>
<li>Address barriers to broader adoption of social impact evaluation of Generative AI systems</li>
<li>Develop policy recommendations for investment in future directions for social impact evaluations</li>
<li>Create a framework for documenting and standardizing evaluation practices</li>
</ol>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<h2 class="title is-3">Call for Tiny Papers</h2>
<div class="content has-text-justified">
<p>We are soliciting tiny papers (up to 2 pages long) in the following formats:</p>
<ol>
<li>Extended Abstracts: Short but complete research papers presenting original or interesting results around social impact evaluation for generative AI.</li>
<li>"Provocations": Novel perspectives or challenges to conventional wisdom around social impact evaluation for generative AI.</li>
</ol>
<h3 class="title is-4">Themes for Submissions</h3>
<p>We welcome submissions addressing, but not limited to, the following themes:</p>
<ol>
<li>Conceptualization and operationalization issues in evaluations of:
<ul>
<li>Bias, stereotypes, and representational harms</li>
<li>Cultural values and sensitive content</li>
<li>Community-centered definitions of disparate performance and privacy</li>
<li>Documentation frameworks for financial and environmental costs of evaluations</li>
</ul>
</li>
<li>Ethical or consequential validity considerations for:
<ul>
<li>Data protection</li>
<li>Data and content moderation labor</li>
<li>Historical implications of evaluation data or practices for evaluation validity</li>
</ul>
</li>
<li>Interrogating or critiquing the theoretical basis of existing evaluations</li>
<li>Novel methodologies for evaluating social impact across different AI modalities</li>
<li>Comparative analyses of existing evaluation frameworks and their effectiveness</li>
<li>Case studies of social impact evaluations in real-world AI applications</li>
</ol>
<h3 class="title is-4">Submission Guidelines</h3>
<ul>
<li>Paper Length: Maximum 2 pages, excluding an unlimited amount of references</li>
<li>Format: PDF file, using the <a href="#" target="_blank">NeurIPS 2024 LaTeX style file</a></li>
<li>Submission Portal: [Insert submission portal link here]</li>
<li>Anonymity: Submissions should be anonymous for two-way anonymized review.</li>
<li>This is a participatory, in-person event. Accepted Authors are encouraged to present their work and discuss it at the event.</li>
<li>Broader impact statement and Limitation section are not counted in the paper length.</li>
</ul>
<h3 class="title is-4">Important Dates</h3>
<ul>
<li>Submission Deadline: August 1, 2024</li>
<li>Notification of Acceptance: September 1, 2024</li>
<li>Workshop Date: [Insert workshop date here]</li>
</ul>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<h2 class="title is-3">Workshop Structure</h2>
<div class="content">
<p>Total Duration: 8 Hours</p>
<table class="table is-fullwidth">
<thead>
<tr>
<th>Time</th>
<th>Session</th>
<th>Description</th>
</tr>
</thead>
<tbody>
<tr>
<td>9:00 AM - 9:30 AM</td>
<td>Welcome and Introduction</td>
<td>
<ul>
<li>Opening remarks</li>
<li>Overview of workshop structure and objectives</li>
</ul>
</td>
</tr>
<tr>
<td>9:30 AM - 11:00 AM</td>
<td>Reflections on the Landscape</td>
<td>
<ul>
<li>Collaborative reflection on the existing landscape</li>
<li>Talks, panels, and breakouts by modality (text, images, audio, video, and multimodal data)</li>
<li>Topics: Underlying frameworks, Contextualization challenges, Defining robust evaluations, Incentive structures</li>
</ul>
</td>
</tr>
<tr>
<td>11:00 AM - 11:15 AM</td>
<td>Break</td>
<td></td>
</tr>
<tr>
<td>11:15 AM - 12:45 PM</td>
<td>Talks + Provocations</td>
<td>
<ul>
<li>Invited speakers to present on current technical evaluations for base models across all modalities</li>
<li>Key social impact categories covered: Bias and stereotyping, Cultural values, Performance disparities, Privacy, Financial and environmental costs, Data moderator labor</li>
<li>Presentations of accepted provocations</li>
</ul>
</td>
</tr>
<tr>
<td>12:45 PM - 1:45 PM</td>
<td>Lunch Break</td>
<td></td>
</tr>
<tr>
<td>1:45 PM - 3:45 PM</td>
<td>Group Activity</td>
<td>
<ul>
<li>Participants break into groups focusing on key social impact categories</li>
<li>Activities include: Choosing Evaluations, Reviewing Tools and Datasets, Examining construct reliability, validity, and ranking methodologies</li>
</ul>
</td>
</tr>
<tr>
<td>3:45 PM - 4:00 PM</td>
<td>Break</td>
<td></td>
</tr>
<tr>
<td>4:00 PM - 5:45 PM</td>
<td>What's Next? Documentation + Resources</td>
<td>
<ul>
<li>Develop policy guidance highlighting impact categories, subcategories, and modalities requiring further investment</li>
<li>Discussions on: Documenting Methods, Developing Shareable Resources, Underlying Frameworks, Contextualization Challenges, Defining Robust Evaluations</li>
</ul>
</td>
</tr>
<tr>
<td>5:45 PM - 6:00 PM</td>
<td>Closing Remarks</td>
<td></td>
</tr>
</tbody>
</table>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<h2 class="title is-3">Invited Speakers</h2>
<div class="content">
<h3 class="title is-4">Confirmed Speakers:</h3>
<ol>
<li>
<strong>Abigail Jacobs</strong>
<ul>
<li>Assistant Professor, School of Information</li>
<li>Assistant Professor of Complex Systems, College of Literature, Science, and the Arts</li>
<li>University of Michigan</li>
</ul>
</li>
<li>
<strong>Nitarshan Rajkumar</strong>
<ul>
<li>Cofounder of UK AI Safety Institute</li>
<li>Adviser to the Secretary of State of the UK Department for Science, Innovation and Technology</li>
</ul>
</li>
<li>
<strong>Su Lin Blodgett</strong>
<ul>
<li>Senior Researcher, Microsoft Research Montreal</li>
</ul>
</li>
</ol>
<h3 class="title is-4">Tentative Speaker:</h3>
<ol start="4">
<li>
<strong>Abeba Birhane</strong>
<ul>
<li>Adjunct Lecturer/Assistant Professor, Trinity College Dublin</li>
<li>Senior Fellow in Trustworthy AI at Mozilla Foundation</li>
</ul>
</li>
</ol>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<h2 class="title is-3">Expected Outcomes</h2>
<div class="content has-text-justified">
<p>Three months after the workshop, we aim to achieve the following outcomes:</p>
<ol>
<li>
<strong>Evaluation Report and Resources/Repository:</strong>
<ul>
<li>Publish a comprehensive summary of the workshop findings</li>
<li>Update resources including:
<ul>
<li>Documentation framework for standardizing evaluation practices</li>
<li>Open source repository addressing identified barriers to broader adoption of social impact evaluation of Generative AI systems</li>
</ul>
</li>
</ul>
</li>
<li>
<strong>Policy Recommendations:</strong>
<ul>
<li>Share detailed policy recommendations for investment in future directions for social impact evaluations based on group discussions and workshop outcomes</li>
</ul>
</li>
<li>
<strong>Knowledge Sharing:</strong>
<ul>
<li>Foster a more systematic and effective approach to evaluating the social impact of generative AI systems by disseminating lessons and findings to the broader AI research community</li>
</ul>
</li>
</ol>
</div>
</div>
</section>
<section class="section"> |