Skip to content

Commit e73bb06

Browse files
jfpariejfparie
authored
Merge pull request #400 from NGO-Algorithm-Audit/feature/structural_edits
EN NL > new subdomain technical-tools/eval-gen-ai
2 parents 206bbef + bada0ad commit e73bb06

13 files changed

Lines changed: 123 additions & 7 deletions

File tree

‎config/_default/menus.NL.toml‎

Lines changed: 8 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -110,11 +110,18 @@ url = "/nl/technical-tools"
110110
icon = "fa-table"
111111
[[main]]
112112
parent = "Technische tools"
113-
name = "AI AQT"
113+
name = "AI en algoritmes kwalificatie tool (AI AQT)"
114114
url = "/nl/technical-tools/AI-AQT"
115115
weight = 4
116116
[[main.params]]
117117
icon = "fa-file"
118+
[[main]]
119+
parent = "Technische tools"
120+
name = "Evaluatie generatieve AI"
121+
url = "/nl/technical-tools/eval-gen-ai"
122+
weight = 5
123+
[[main.params]]
124+
icon = "fa-robot"
118125

119126
[[main]]
120127
name = "Evenementen"

‎config/_default/menus.en.toml‎

Lines changed: 8 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -110,11 +110,18 @@ url = "/technical-tools"
110110
icon = "fa-table"
111111
[[main]]
112112
parent = "Technical tools"
113-
name = "AI and Algorithms Qualification Toolkit (AI AQT)"
113+
name = "AI and Algorithms Qualification Tool (AI AQT)"
114114
url = "/technical-tools/implementation-tool"
115115
weight = 4
116116
[[main.params]]
117117
icon = "fa-file"
118+
[[main]]
119+
parent = "Technical tools"
120+
name = "Evaluating generative AI"
121+
url = "/technical-tools/eval-gen-ai"
122+
weight = 5
123+
[[main.params]]
124+
icon = "fa-robot"
118125

119126
[[main]]
120127
name = "Events"

‎content/.DS_Store‎

0 Bytes
Binary file not shown.

‎content/english/_index.md‎

Lines changed: 2 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -91,9 +91,8 @@ Areas_of_AI_expertise:
9191
Evaluating Large Language Models (LLMs) and other general-purpose AI
9292
models for robustness, privacy and AI Act compliance. Based on
9393
real-world examples, are developing a framework to analyze content
94-
filters, guardrails and user interaction design choices. <a
95-
href="/knowledge-platform/project-work/#LLM-validation"
96-
style="text-decoration: underline;">Learn more</a> about our evaluation
94+
filters, guardrails and user interaction design choices. <a href="/technical-tools/eval-gen-ai" style="text-decoration:
95+
underline;">Learn more</a> about our evaluation
9796
framework.
9897
- name: AI Act implementation and standards
9998
icon: fas fa-certificate

‎content/english/technical-tools/eval-gen-ai.md‎

Lines changed: 52 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,52 @@
1+
---
2+
type: regular
3+
title: Evaluation of generative AI
4+
subtitle: >
5+
Overview of Algorithm Audit's expertise pertaining to generative AI.
6+
image: /images/technical tools/eval-gen-ai/Kader.png
7+
quick_navigation:
8+
title: Content overview
9+
links:
10+
- title: Introduction
11+
url: '#intro'
12+
- title: Validation framework
13+
url: '#validation-framework'
14+
- title: Development and usage
15+
url: '#development-and-usage'
16+
promo_bar:
17+
- content: "**\U0001F44B Do you want to learn more about Algorithm Audit's expertise on generative AI? Get in [contact](/en/about/contact).**"
18+
---
19+
20+
{{< promo_bar index="0" >}}
21+
22+
<!-- Introductie -->
23+
24+
{{< container_open icon="far fa-file" title="Introduction" id="intro" >}}
25+
26+
Algorithm Audit has versatile expertise in evaluating generative AI. We have contributed to developing a validation framework for responsible use of large language models (LLMs) for public information provision. Also, we conduct project work for the AI Office of the European Commission to evaluate socio-technical risks of general purpose AI (GPAI) models.
27+
28+
{{< container_close >}}
29+
30+
31+
<!-- Validation framework -->
32+
33+
{{< container_open icon="far fa-check" title="Validation framework" id="validation-framework" >}}
34+
35+
Soon available in English.
36+
37+
<!-- {{< embed_pdf url="/pdf-files/technical-tools/UBDT/20260215 Auditing a Dutch Public Sector Risk Profiling Algorithm.pdf" width_mobile_pdf="12" width_desktop_pdf="12" >}} -->
38+
39+
{{< container_close >}}
40+
41+
42+
<!-- Development and usage -->
43+
44+
{{< container_open icon="fas fa-terminal" title="Development and usage of validation framework:" id="development-and-usage" >}}
45+
46+
The validation framework 'Responsible use of Large Language Models (LLMs) for public information provision' has been developed in collaboration with the Dutch judiciary, Technical University Eindhoven (TU/e), T&T Data Consultancy and Deloitte. The validation framework is for a large extent inspired on the LLM-pilot [voorrecht-Rechtspraak](https://www.voorrecht-rechtspraak.nl). The project is supported by the program <a href="https://www.sidnfonds.nl/projecten/validatiekader-betrouwbare-llms-voor-publieke-informatievoorziening" target="_blank">‘Responsible AI in de praktijk'</a> of the SIDN Fund and TopSectorICT.
47+
48+
The Validation framework is published under the CC BY-4.0 license.
49+
50+
{{< image image1="/images/partner%20logo-cropped/Rechtspraak.png" image2="/images/partner%20logo-cropped/TUEindhoven.svg.png" image3="/images/partner%20logo-cropped/T&TDataConsultancy.png" image4="/images/partner%20logo-cropped/Deloitte.png" width_desktop="3" width_mobile="1" alt1="De Rechtspraak" alt2="Technische Universiteit Eindhoven" alt3="T&T Data Consultancy" alt4="Deloitte" caption1="De Rechtspraak" caption2="Technische Universiteit Eindhoven" caption3="T&T Data Consultancy" caption4="Deloitte" >}}
51+
52+
{{< container_close >}}

‎content/nederlands/_index.md‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -97,7 +97,7 @@ Areas_of_AI_expertise:
9797
validatiekader om contentfilters, guardrails en ontwerpkeuzes voor
9898
gebruikersinteractie te beoordelen. <a
9999
100-
href="/nl/knowledge-platform/project-work/#LLM-validation"
100+
href="/nl/technical-tools/eval-gen-ai"
101101
102102
style="text-decoration: underline;">Lees meer</a> over ons
103103
validatiekader.

‎content/nederlands/technical-tools/eval-gen-ai.md‎

Lines changed: 51 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,51 @@
1+
---
2+
type: regular
3+
title: Evaluatie van generatieve AI
4+
subtitle: >
5+
Overzicht van Algorithm Audit's expertise met betrekking tot generatieve AI.
6+
image: /images/technical tools/eval-gen-ai/Kader.png
7+
quick_navigation:
8+
title: Inhoudsopgave
9+
links:
10+
- title: Introductie
11+
url: '#intro'
12+
- title: Validatiekader
13+
url: '#validation-framework'
14+
- title: Ontwikkeling en gebruik
15+
url: '#development-and-usage'
16+
promo_bar:
17+
- content: "**\U0001F44B Wil je meer weten over Algorithm Audit's expertise over generatieve AI? Stuur een [bericht](/nl/about/contact).**"
18+
---
19+
20+
{{< promo_bar index="0" >}}
21+
22+
<!-- Introductie -->
23+
24+
{{< container_open icon="far fa-file" title="Introductie" id="intro" >}}
25+
26+
Algorithm Audit heeft diverse expertise over het evalueren van generatieve AI. We hebben bijgedragen aan de ontwikkeling van een validatiekader voor het verantwoord gebruik van large language models (LLM’s) voor publieke informatievoorziening. Daarnaast verrichten we projectwerk voor de AI Office van de Europese Commissie om de sociotechnische risico’s van AI modellen voor algemene doeleinden (GPAI) te evalueren.
27+
28+
{{< container_close >}}
29+
30+
31+
<!-- Validatiekader -->
32+
33+
{{< container_open icon="far fa-check" title="Validatiekader" id="validation-framework" >}}
34+
35+
Wordt binnenkort gepubliceerd
36+
37+
<!-- {{< embed_pdf url="/pdf-files/technical-tools/UBDT/20260215 Auditing a Dutch Public Sector Risk Profiling Algorithm.pdf" width_mobile_pdf="12" width_desktop_pdf="12" >}} -->
38+
39+
{{< container_close >}}
40+
41+
42+
<!-- Ontwikkeling en gebruik -->
43+
44+
{{< container_open icon="fas fa-terminal" title="Ontwikkeling en gebruik van validatiekader:" id="development-and-usage" >}}
45+
Het validatiekader 'Verantwoord gebruik van Large Language Models (LLM's) voor publieke informatievoorziening' is ontwikkeld in samenwerking met de Rechtspraak, de Technische Universiteit Eindhoven (TU/e), T&T Data Consultancy en Deloitte. Het validatiekader is voor een groot deel gebaseerd op de LLM-pilot <a href="https://www.voorrecht-rechtspraak.nl" target="_blank">voorrecht-Rechtspraak</a>. Het project wordt ondersteund door het programma <a href="https://www.sidnfonds.nl/projecten/validatiekader-betrouwbare-llms-voor-publieke-informatievoorziening" target="_blank">'Responsible AI in de praktijk'</a> van het SIDN Fonds en TopSectorICT.
46+
47+
Het validatiekader is gepubliceerd onder de CC BY-4.0 licentie.
48+
49+
{{< image image1="/images/partner%20logo-cropped/Rechtspraak.png" image2="/images/partner%20logo-cropped/TUEindhoven.svg.png" image3="/images/partner%20logo-cropped/T&TDataConsultancy.png" image4="/images/partner%20logo-cropped/Deloitte.png" width_desktop="3" width_mobile="1" alt1="De Rechtspraak" alt2="Technische Universiteit Eindhoven" alt3="T&T Data Consultancy" alt4="Deloitte" caption1="De Rechtspraak" caption2="Technische Universiteit Eindhoven" caption3="T&T Data Consultancy" caption4="Deloitte" >}}
50+
51+
{{< container_close >}}

‎content/nederlands/technical-tools/implementation-tool.md‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
---
22
type: regular
3-
title: AI en algoritmes kwalificatietoolkit (AI AQT)
3+
title: AI en algoritmes kwalificatie tool (AI AQT)
44
subtitle: >
55
AI AQT is een tool die ondersteunt bij het identificeren en
66
risicoclassificeren van AI en andere data-gedreven systemen. Complexe

‎static/images/partner logo-cropped/Deloitte.png‎

53.9 KB
Loading

‎static/images/partner logo-cropped/Rechtspraak.png‎

98.4 KB
Loading

0 commit comments

Comments
 (0)