Responible and Responsive: International Decision-Making in AI Underlag till styrelsemöte 1-2 april 2007

Till: Styrelsen (Dokument fr AI DOC)
Datum: 4 april 2007 (Urspr. 18 april 2006)

Responsible and Responsive: International Decision-Making in AI

AI Index: ORG 21/001/2006 Distr: No. of words: 1,678 Peter Benenson House
Amnesty International
International Secretariat
1 Easton Street
London WC1X 0DW

Responsible and Responsive:
International Decision-Making in AI

Date: 18 April 2006

This is a report of the meeting on decision-making in Amnesty International held in London 4-6 March 2006.

The meeting was called by the IEC to generate advice and recommendations on the future of AI’s decision-making. A careful analysis of AI’s current position on decision-making identified the external context of AI’s work, and how its decision-making processes could be developed to better respond to human rights violations. Key principles of best practice in decision-making, on which leadership should be based, were outlined. AI needs to develop new, revolutionary approaches to its global decision-making, replacing the current model with a bold, fast and flexible architecture.

Areas of attention identified included a need for a deeper diversity in AI’s decision-making input, a need to strengthen AI’s accountability at many levels within the movement, the importance of giving leadership the flexibility and freedom to make decisions without constantly seeking permission before the fact, and the possibilities offered by globalizing AI’s leadership and decision-making.

A goal of achieving a culture of responsive and trusted leadership in AI was also proposed, and a number of specific recommendations made to the IEC.

Meeting participants: Soledad Garcia Munoz, Ian Gibson, Kate Gilmore, Lilian Goncalves, Christine Pamp, Guadelupe Rivas, Deborah Smith, Gaby Smyth, Bart Stapert, Russell Thirgood, Brian Dooley.

Apologies: Paul Hoffman, Irene Khan, Carlos Muniz,

decision making, accountability, leadership


This is an internal circular which is being sent to all sections and coordinating structures.

Recommended actions

Please circulate this document to all people in your section/structure who are involved in decision-making.


Concerns about the future ability of Amnesty International to make maximum positive impact for human rights are directly linked to our concerns about its future approach to decision-making. The International Council Meeting (ICM), the Chairs’ Forum, the International Secretariat and sections and structures have been searching for a way to make decision-making more effective and efficient, especially following the introduction of a three-year cycle for ICMs after 2009.

The agreed purpose of the meeting was to explore and advise the IEC on future directions for a decision-making in AI, beginning with international decision-making institutions, but then looking at how all AI’s decision-making serves AI’s mission in terms of human rights effectiveness.

The meeting proceeded through a series of workshops and discussions analyzing AI’s overall situation, looking at what were the most important features of decision-making for AI in the future, identifying themes and priorities, and finally thinking about next steps and communication to the movement.

Situational Analysis

People’s human rights are and will be under attack in new and old ways by new and old actors. People are passionate about their human rights and are searching for language to describe these rights and the violation of them. Meanwhile, the relevance of human rights discourse is not well established everywhere, and what makes for relevant, inclusive and empowering discourse is under contest.

Civil society has become a complex space, with greater competition for public attention. There is also a higher expectation of co-operation and partnerships between NGOs, and greater scrutiny and scepticism of NGOs. NGOs need to pay greater attention to risk management, and offer greater transparency to the public and its peer organizations.

For AI, the landscape is challenging as we are neither diverse enough nor reflective of the societies we work in or for. The number of people participating in decision-making in AI is very small. It is not unusual in medium and large sections for the proportion of AI members attending AGMs, electing the board or otherwise actively participating in decision-making to be less than one per cent. Nevertheless, the external human rights environment offers AI significant opportunities, and there is a feeling that this can be AI’s moment if it can respond ambitiously, quickly and effectively.

Desirable Decision-Making Processes

Given the external world challenges facing AI, internal decision-making in the future ought to be accountable, rapid and responsive, active and visible, genuinely having an impact, change-oriented, and based on deep diversity. There is a tension between lengthy consultation, as a basis for good decision-making, and AI’s need for quick, quality decisions. One of the key issues here is shifting attention and energy from the work that precedes a decision to subsequent accountability for how the decision was taken, what has been done to implement it, and what have been the consequences of its implementation.

Key principles on which AI leadership ought to be based include:

  • AI being led by people who are skilled in leadership;
  • AI leaders trusting themselves and others;
  • AI leaders being honest about their actions and ready to take responsibility for the consequences of those actions;
  • AI leaders being less constrained by rules;
  • AI leaders bringing in the voice of beneficiaries to balance the input on decision-making.


AI has been talking about diversity for many years, and the issue remains an obvious problem both within sections and structures and in the worldwide movement.

The challenges involved in AI using its existing supporter base to develop deep diversity include AI not being easy enough to find or join and AI not being able to reach others to join. To create real diversity, it is not enough just to recruit diversity, but the psychology and culture of AI will have to change to attract and retain a more diverse supporter base.

Finding mechanisms to encourage input for those with and for whom AI works—beneficiaries and potential beneficiaries—is a key component of develop AI’s diversity. It is also vital for AI to develop local relevance throughout the world.


The concept of tough trust needs to be rooted in stronger accountability, and AI needs to develop robust methods for this. ICM resolutions reappear again and again because the IEC, IS and sections and structures do not do what the resolutions ask. The IEC and IS accountability sessions at the last ICM, although an improvement, were unsatisfactory, as was the session on section and structure accountability. Would AI see it as being desirable (or possible) to see "accountability" as being having leaders held accountable by those who are led and having those who are led also held accountable by the leadership?

A trusted system for accountability must be developed across the movement. AI should move quickly to establish a system of ‘fast trust,’ recognising that taking responsibility for decisions after the act is quicker than seeking permission before the fact. Accountability throughout the movement will depend on honesty, trust and frankness in everyone’s communications.


AI should examine how its international leaders are currently identified, selected, assessed, supported, appraised and held accountable. It needs to develop a leadership that is trustworthy, courageous and trusted, and needs to create safe space for unpractised leaders to practise and develop organically.

Section and structure representatives at international decision-making meetings should enjoy greater trust and flexibility, and be empowered to use their own judgement rather than having to refer all decisions back to sections and structures.

Leadership ought to effectively delegate power, be based on co-ordination and not control, and enjoy greater freedom in decision-making.

Globalizing AI’s Decision-Making

AI has developed systems for globalizing governance, but not for globalizing operations. AI’s economy is being globalized but its human resources and expertise are not.

If we were to move to a global model of decision-making, resources (including financial) might be pooled towards a common goal of greater positive impact for human rights. Leaders might have to step out of the way to allow those with expertise the space to act and then hold them accountable to agreed goals afterwards.

Such a radical shift in AI culture would have to overcome the resistance to change the ways we have always done things, trust people to use resources wisely and introduce innovative techniques in decision-making.

"The words"

When those present came to identify the words that best captured and summarized their vision for what represented their vision for leadership in AI, they produced this list:

Trust, mutual respect and self respect, humility; using all available forms of communication, diversity, passion, closer to the "victims", impact/evaluation, expertise, humanism, freedom, explicit leadership, maximizing impact.


The meeting achieved consensus on the conclusions and on its recommendations to the IEC, and ended with a high degree of enthusiasm about these conclusions as a way forward for decision-making in AI. It identified as the next steps for the IEC:

  • Igniting a discussion among supporters within sections and structure about how to incorporate these insights into AI’s future decision-making;
  • Communicate with the movement on new approaches to decision-making, suggesting a short paper be sent out asking what globalizing decision-making would mean for various parts of AI;
  • A discussion to encourage the Chairs’ Forum to undertake exercises on decision-making similar to those at this meeting, and consider the outcomes of their discussions;
  • Explore creating an electronic open space for AI people to meet and share ideas – a sort of intranet for the global movement.
  • Explore ways of enabling and encouraging the voice of those AI works with and for to balance the input into decision-making;
  • Task someone to draft a roadmap for specific actions arising from the work of this meeting;
  • Identify barriers to change, and how they might be overcome

Papers Provided to the Meeting

A summary of DAWG’s evaluation of the current decision-making structures of AI ORG21/06/2000

What does democracy mean? A resource paper on Decision-Making ORG21/07/2000

Decision-making in other organizations. A resources paper on decision-making ORG21/08/2000

Changing the way we change: Options for reforming AI’s decision-making processes and structures ORG21/10/2000

International decision-making in the future ORG 50/014/2001

Policy-setting paper POL10/003/04

Section comments on policy-setting paper

Globalizing AI ORG 30/011/2005

Statute of Amnesty International in 1999 Part IV of ORG 52/001/2000

Some ICM decisions relevant to this project

List of ICM decisions from 2001, 2003 and 2005

2003 and 2005 ICMs


Amnesty International, International Secretariat, 1 Easton Street, WC1X 0DW, London, United Kingdom