-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy path1.2_responsibility_part2.html
More file actions
514 lines (439 loc) · 25.3 KB
/
1.2_responsibility_part2.html
File metadata and controls
514 lines (439 loc) · 25.3 KB
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
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>1.2 Exposure to toxic content - Responsibility</title>
<link href="https://fonts.googleapis.com/css2?family=Figtree:wght@300;400;500;600;700&display=swap" rel="stylesheet">
<style>
* {
margin: 0;
padding: 0;
box-sizing: border-box;
}
body {
font-family: 'Figtree', -apple-system, BlinkMacSystemFont, 'Segoe UI', sans-serif;
background-color: #ffffff;
color: #000000;
line-height: 1.3;
}
.container {
max-width: 1200px;
margin: 0 auto;
padding: 8px;
flex: 1;
min-width: 200px;
overflow-wrap: break-word;
word-break: break-word; }
h1 {
text-align: center;
margin-bottom: 8px;
color: #000000;
font-weight: 600;
font-size: 18px;
}
.legend {
text-align: center;
font-size: 12px;
color: #888888;
font-style: italic;
margin-bottom: 12px;
padding: 8px;
background-color: #f9f9f9;
border-radius: 5px;
border: 1px solid #e0e0e0;
}
.selection-title {
text-align: center;
font-size: 14px;
font-weight: 600;
color: #666666;
margin-bottom: 10px;
}
.nav-pills {
display: flex;
flex-wrap: wrap;
gap: 4px;
margin-bottom: 15px;
justify-content: center;
}
.nav-pill {
background: #f8f9fa;
border: 1px solid #e0e0e0;
border-radius: 25px;
padding: 12px 20px;
cursor: pointer;
font-family: 'Figtree', sans-serif;
font-size: 16px;
font-weight: 500;
transition: all 0.3s ease;
color: #000000;
}
.nav-pill:hover {
background: #e9ecef;
border-color: #000000;
}
.nav-pill.active {
background: #000000;
color: white;
border-color: #000000;
}
.actor-section {
display: none;
}
.actor-section.active {
display: block;
}
.content-grid {
display: flex;
width: 100%;
gap: 4px;
}
.content-column {
background: #ffffff;
border: 1px solid #e0e0e0;
border-radius: 8px;
padding: 8px;
flex: 1;
min-width: 200px;
overflow-wrap: break-word;
word-break: break-word; }
.criteria-header {
font-size: 12px;
font-weight: 600;
margin-bottom: 15px;
padding-bottom: 10px;
border-bottom: 2px solid;
}
.criteria-header.higher {
color: #FF0000;
border-bottom-color: #FF0000;
}
.criteria-header.lower {
color: #2E5C8A;
border-bottom-color: #2E5C8A;
}
.summary-section {
margin-bottom: 20px;
}
.summary-text {
margin-bottom: 15px;
font-weight: 500;
color: #000000;
font-size: 15px;
}
.quote-details {
margin-top: 15px;
}
.quote-toggle {
cursor: pointer;
color: #000000;
font-weight: 500;
font-size: 16px;
background-color: #ffff00;
padding: 10px 15px;
border-radius: 4px;
display: inline-block;
}
.quote-toggle:hover {
color: #333333;
}
.quote-list {
margin-top: 15px;
padding-left: 20px;
}
.quote-list li {
margin-bottom: 12px;
font-size: 16px;
padding: 10px 15px;
line-height: 1.3;
color: #000000;
}
@media (max-width: 768px) {
.content-grid {
gap: 4px;
}
.selection-title {
text-align: center;
font-size: 14px;
font-weight: 600;
color: #666666;
margin-bottom: 10px;
}
.nav-pills {
justify-content: flex-start;
}
.nav-pill {
font-size: 16px;
padding: 4px 8px;
}
}
</style>
</head>
<body>
<div class="container">
<h1>1.2 Exposure to toxic content - Responsibility</h1>
<div class="selection-title">Select an actor:</div>
<div class="nav-pills">
<button class="nav-pill active" data-target="AIDeveloperSpecializedAI">
AI Developer (Specialized AI)
</button>
<button class="nav-pill" data-target="AIInfrastructureProvider">
AI Infrastructure Provider
</button>
<button class="nav-pill" data-target="AffectedStakeholder">
Affected Stakeholder
</button>
</div>
<div class="content-sections">
<div class="actor-section active" id="AIDeveloperSpecializedAI">
<div class="content-grid">
<div class="content-column">
<h3 class="criteria-header higher">Reasons for Higher Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> Respondents emphasized primary responsibility because developers choose training data, model architecture, alignment techniques, and post-training filters that directly determine whether models emit toxic content. They are usually the only actors with full technical insight into model internals, and their choices about processing and modeling data—including training on unmoderated internet sources and poor filtering of NSFW or extremist datasets—directly enable toxic content generation.</p>
<details class="quote-details">
<summary class="quote-toggle">See all expert comments (2)</summary>
<ul class="quote-list">
<li>"General-purpose and specialized AI developers are primarily responsible because they choose, process, and model the data that can carry or amplify toxic content. This includes training on unmoderated internet sources, poor filtering of NSFW or extremist datasets, and failure to build semantic toxicity detection into system architecture."</li> <li>"AI Developer (specialized AI): They choose the training data, model architecture, alignment techniques and post-training filters that most directly determine whether the model emits toxic content. They are also usually the only actors with full technical insight into model internals."</li>
</ul>
</details>
</div>
</div>
<div class="content-column">
<h3 class="criteria-header lower">Reasons for Lower Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> [NO EXPERT COMMENTS PROVIDED]</p>
</div>
</div>
</div>
</div>
<div class="actor-section" id="AIInfrastructureProvider">
<div class="content-grid">
<div class="content-column">
<h3 class="criteria-header higher">Reasons for Higher Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> [NO EXPERT COMMENTS PROVIDED]</p>
</div>
</div>
<div class="content-column">
<h3 class="criteria-header lower">Reasons for Lower Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> One expert commented: "Infrastructure providers were rated minimally responsible because their role is passive in relation to content. However, as scale-enablers, they may bear more responsibility in the future if content toxicity becomes a service-level issue."</p>
<details class="quote-details">
<summary class="quote-toggle">See all expert comments (1)</summary>
<ul class="quote-list">
<li>"Infrastructure providers were rated minimally responsible because their role is passive in relation to content. However, as scale-enablers, they may bear more responsibility in the future if content toxicity becomes a service-level issue."</li>
</ul>
</details>
</div>
</div>
</div>
</div>
<div class="actor-section" id="AffectedStakeholder">
<div class="content-grid">
<div class="content-column">
<h3 class="criteria-header higher">Reasons for Higher Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> [SUMMARY TBC]</p>
<details class="quote-details">
<summary class="quote-toggle">See all expert comments (2)</summary>
<ul class="quote-list">
<li>"I did not change my assessment even though I was an outlier on how much affected stakeholders are responsible. I think they do have a responsibility to monitor usage of AI by e.g. their employees and they are responsible for checking content etc."</li> <li>"One respondent maintained that affected stakeholders have responsibility to monitor AI usage by employees and check content, despite being an outlier in this assessment."</li>
</ul>
</details>
</div>
</div>
<div class="content-column">
<h3 class="criteria-header lower">Reasons for Lower Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> [SUMMARY TBC]</p>
<details class="quote-details">
<summary class="quote-toggle">See all expert comments (2)</summary>
<ul class="quote-list">
<li>"Users and affected stakeholders are not responsible. They do not shape, publish, or promote the toxic material - they experience its consequences, often without visibility or consent."</li> <li>"Respondents argued affected stakeholders are not responsible because they do not shape, publish, or promote toxic material—they experience its consequences, often without visibility or consent."</li>
</ul>
</details>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
<div class="actor-section" id="AIDeveloperSpecializedAI">
<div class="content-grid">
<div class="content-column">
<h3 class="criteria-header higher">Reasons for Higher Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> Respondents emphasized primary responsibility because developers choose training data, model architecture, alignment techniques, and post-training filters that directly determine whether models emit toxic content. They are usually the only actors with full technical insight into model internals, and their choices about processing and modeling data—including training on unmoderated internet sources and poor filtering of NSFW or extremist datasets—directly enable toxic content generation.</p>
<details class="quote-details">
<summary class="quote-toggle">See all expert comments (2)</summary>
<ul class="quote-list">
<li>"General-purpose and specialized AI developers are primarily responsible because they choose, process, and model the data that can carry or amplify toxic content. This includes training on unmoderated internet sources, poor filtering of NSFW or extremist datasets, and failure to build semantic toxicity detection into system architecture."</li> <li>"AI Developer (specialized AI): They choose the training data, model architecture, alignment techniques and post-training filters that most directly determine whether the model emits toxic content. They are also usually the only actors with full technical insight into model internals."</li>
</ul>
</details>
</div>
</div>
<div class="content-column">
<h3 class="criteria-header lower">Reasons for Lower Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> [NO EXPERT COMMENTS PROVIDED]</p>
</div>
</div>
</div>
</div>
<div class="actor-section" id="AIDeployer">
<div class="content-grid">
<div class="content-column">
<h3 class="criteria-header higher">Reasons for Higher Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> [SUMMARY TBC]</p>
<details class="quote-details">
<summary class="quote-toggle">See all expert comments (3)</summary>
<ul class="quote-list">
<li>"Deployers and governance actors are highly responsible. Deployers distribute models without proper content controls or transparency about what may emerge. Governance actors fail to enforce platform accountability or moderation requirements - especially for generative models serving vulnerable populations."</li> <li>"My one disagreement with the consensus is that I think developers and deployers are both highly responsible, which is where I rated them equally rather than one having primary responsibility. Developers absolutely have to do their due diligence, but deployers also have special responsibility to vet tools including betting developers themselves and then put tools in front of users. These are highly interrelated relationships, so assigning primary responsibility does not make sense to me."</li> <li>"Respondents noted deployers are highly responsible because they distribute models without proper content controls or transparency about what may emerge, and they have special responsibility to vet tools and developers before putting tools in front of users. Some argued deployers share equal responsibility with developers rather than one having primary responsibility, given the highly interrelated nature of these relationships."</li>
</ul>
</details>
</div>
</div>
<div class="content-column">
<h3 class="criteria-header lower">Reasons for Lower Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> [SUMMARY TBC]</p>
<details class="quote-details">
<summary class="quote-toggle">See all expert comments (1)</summary>
<ul class="quote-list">
<li>"[NO EXPERT COMMENTS PROVIDED]"</li>
</ul>
</details>
</div>
</div>
</div>
</div>
<div class="actor-section" id="AIInfrastructureProvider">
<div class="content-grid">
<div class="content-column">
<h3 class="criteria-header higher">Reasons for Higher Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> [NO EXPERT COMMENTS PROVIDED]</p>
</div>
</div>
<div class="content-column">
<h3 class="criteria-header lower">Reasons for Lower Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> One expert commented: "Infrastructure providers were rated minimally responsible because their role is passive in relation to content. However, as scale-enablers, they may bear more responsibility in the future if content toxicity becomes a service-level issue."</p>
<details class="quote-details">
<summary class="quote-toggle">See all expert comments (1)</summary>
<ul class="quote-list">
<li>"Infrastructure providers were rated minimally responsible because their role is passive in relation to content. However, as scale-enablers, they may bear more responsibility in the future if content toxicity becomes a service-level issue."</li>
</ul>
</details>
</div>
</div>
</div>
</div>
<div class="actor-section" id="AIUser">
<div class="content-grid">
<div class="content-column">
<h3 class="criteria-header higher">Reasons for Higher Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> Multiple respondents argued users are primarily responsible because most toxic content will be created intentionally by AI users—including CSAM, hate speech, and pornography. Users proactively create toxic content and try to find new ways to jailbreak guardrails, making it strange not to hold them primarily responsible when they do this.</p>
<details class="quote-details">
<summary class="quote-toggle">See all expert comments (2)</summary>
<ul class="quote-list">
<li>"AI users tend to be the ones proactively creating toxic content and trying to find new ways to jailbreak guardrails. It would be strange not to hold them primarily responsible when they do this."</li> <li>"I think most toxic content, as defined by the study, will be created intentionally by AI users so they are primarily responsible. E.g. CSAM, hate speech, pornography I think is primarily going to be intentional act by the AI user."</li>
</ul>
</details>
</div>
</div>
<div class="content-column">
<h3 class="criteria-header lower">Reasons for Lower Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> One expert commented: "Users and affected stakeholders are not responsible. They do not shape, publish, or promote the toxic material - they experience its consequences, often without visibility or consent."</p>
<details class="quote-details">
<summary class="quote-toggle">See all expert comments (1)</summary>
<ul class="quote-list">
<li>"Users and affected stakeholders are not responsible. They do not shape, publish, or promote the toxic material - they experience its consequences, often without visibility or consent."</li>
</ul>
</details>
</div>
</div>
</div>
</div>
<div class="actor-section" id="AffectedStakeholder">
<div class="content-grid">
<div class="content-column">
<h3 class="criteria-header higher">Reasons for Higher Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> [SUMMARY TBC]</p>
<details class="quote-details">
<summary class="quote-toggle">See all expert comments (2)</summary>
<ul class="quote-list">
<li>"I did not change my assessment even though I was an outlier on how much affected stakeholders are responsible. I think they do have a responsibility to monitor usage of AI by e.g. their employees and they are responsible for checking content etc."</li> <li>"One respondent maintained that affected stakeholders have responsibility to monitor AI usage by employees and check content, despite being an outlier in this assessment."</li>
</ul>
</details>
</div>
</div>
<div class="content-column">
<h3 class="criteria-header lower">Reasons for Lower Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> [SUMMARY TBC]</p>
<details class="quote-details">
<summary class="quote-toggle">See all expert comments (2)</summary>
<ul class="quote-list">
<li>"Users and affected stakeholders are not responsible. They do not shape, publish, or promote the toxic material - they experience its consequences, often without visibility or consent."</li> <li>"Respondents argued affected stakeholders are not responsible because they do not shape, publish, or promote toxic material—they experience its consequences, often without visibility or consent."</li>
</ul>
</details>
</div>
</div>
</div>
</div>
<div class="actor-section" id="AIGovernanceActor">
<div class="content-grid">
<div class="content-column">
<h3 class="criteria-header higher">Reasons for Higher Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> One expert commented: "Deployers and governance actors are highly responsible. Deployers distribute models without proper content controls or transparency about what may emerge. Governance actors fail to enforce platform accountability or moderation requirements - especially for generative models serving vulnerable populations."</p>
<details class="quote-details">
<summary class="quote-toggle">See all expert comments (1)</summary>
<ul class="quote-list">
<li>"Deployers and governance actors are highly responsible. Deployers distribute models without proper content controls or transparency about what may emerge. Governance actors fail to enforce platform accountability or moderation requirements - especially for generative models serving vulnerable populations."</li>
</ul>
</details>
</div>
</div>
<div class="content-column">
<h3 class="criteria-header lower">Reasons for Lower Responsibility</h3>
<div class="summary-section">
<p class="summary-text"><strong>AI-generated summary:</strong> [NO EXPERT COMMENTS PROVIDED]</p>
</div>
</div>
</div>
</div>
</div>
</div>
<script>
document.addEventListener('DOMContentLoaded', function() {
const pills = document.querySelectorAll('.nav-pill');
const sections = document.querySelectorAll('.actor-section');
pills.forEach(pill => {
pill.addEventListener('click', function() {
// Remove active class from all pills and sections
pills.forEach(p => p.classList.remove('active'));
sections.forEach(s => s.classList.remove('active'));
// Add active class to clicked pill
this.classList.add('active');
// Show corresponding section
const targetId = this.getAttribute('data-target');
const targetSection = document.getElementById(targetId);
if (targetSection) {
targetSection.classList.add('active');
}
});
});
});
</script>
</body>
</html>