Hold those who deploy AI capabilities accountable to a diverse forum of stakeholders

Is it possible for a representative sample of society to more directly inform the direction of development of AI capabilities? The task here involves two steps:

• Brainstorm a set of structures which could make diverse stakeholder input more feasible.
  • One possible example could be similar to a council who are compensated to engage with these questions e.g. France's Citizen councils
• Consider whether some structures are more vulnerable to manipulation than others and whether they complement other approaches to reduce the risks from AI development