What do you do if stakeholders in an AI project have conflicting interests?

In any AI project, as with other projects, diverse opinions will emerge. The crucial step is to over-communicate and articulate the shared goals you aim to achieve. Next, pinpoint the areas of disagreement and determine which are vital to the project's success. Agree to disagree on non-critical issues and proceed. Given that this field is still in its early stages, it's natural to encounter differing viewpoints at this point in AI evolution.

Understanding the needs is one of the most challenging aspects of projects. So much emphasis is put on the needs of the client, but it is important that the needs of all the stakeholders should be taken into consideration. Engineers would want to over-engineer what they're making and make it the best version possible, Project managers might want to deliver it as fast as possible, And the client might want to optimize the lowest cost. Enough time should be allocated for the different stakeholders to come together and clarify the requirements of the project and what each party expects out of it. This is at least half the work.

Eines der erfolgskritischsten Teilbereiche entlang der AI Journey ist Zielklarheit. Wir nennen es Solution Framing, um einzugrenzen, was konkret mit KI erreicht werden soll. Dabei geht es darum die unterschiedlichen Stakeholder-Ziele zu diskutieren und untereinander zu synchronisieren. Sobald ein gesamtheitliches Verst?ndnis zu Zielen und Schlüsselergebnissen beim KI Use Case geschaffen wurde, lassen sich die mehrere Ziele in eine Zielhierarchie bringen. Dabei gibt es immer ein prim?res Ziel, das mit hoher Wahrscheinlichkeit den gr??ten Impact durch KI-Einsatz bringt und weitere sekund?re Ziele. In der Praxis sehen wir immer wieder, dass ein gemeinsames Zielverst?ndnis eines der zentralen Erfolgsfaktoren in KI-Projekten ist.

Prioritizing goals is one of the best ways to combat stakeholder misalignment. The goal of deploying an AI solution need to have clear justifications which all stakeholders can agree on. Then as the project is started and moves forward, it will be important to prioritize goals that push the project in the direction to realize the overarching goal.

Es fundamental entender los objetivos del negocio, priorizar lo que realmente va a generar beneficios cuantificables de manera rápida y eficiente debería ser el camino no solo para los proyectos de IA, sino en general. Ahora bien, como contraponemos las prioridades, con datos. Primero los datos asociados al beneficio y segundo los datos asociados a la factibilidad y en ese sentido los proyectos de IA involucran muchos aspectos, no solo el tecnológico sino también la complejidad en la adopción, el impacto en la organización y el impacto hacia el mercado al convertirse en una organización que aprovecha los beneficios de la IA. Todo ello debe ser parte de la ecuación que te permite justificar y priorizar la inversión.

Having a neutral attitude for such conversations always helps! Make sure you're doing a discussion with all the subject matter experts as well as high level management with a goal, a vision and a notepad. Eventually do a job task analysis and that would help chalk out a plan to monitor progress.

Arrange Discussions: Set up forums where stakeholders can openly share their perspectives. Mediate Fairly: Act as a neutral facilitator to ensure balanced participation and understanding. Identify Common Ground: Help stakeholders find shared objectives and concerns as a basis for agreement. Negotiate Solutions: Encourage compromise on conflicting points, focusing on the project's overall success. Document Agreements: Record resolutions and agreed actions to avoid future misunderstandings.

Like it is written in almost any agile books: "Fail fast and often." To be flexible and don't stick with your ideas, when they clearly are not working, is a mindset everybody in the project needs to learn. Especially in AI projects. Since AI is a fairly new domain, which everybody wants to use at the moment, there are so many mistakes you can do. It is necessary to do some of them to learn and anticipate to future attempts. OpenAI was not founded in 2022, but in 2015. They also made a lot of mistakes, but learned from them.

Brainstorm Alternatives: Encourage stakeholders to propose different approaches that could meet their interests. Evaluate Options: Assess the feasibility, benefits, and drawbacks of each alternative, focusing on mutual gains. Redefine Goals: If necessary, adjust the project's objectives to accommodate critical stakeholder needs without compromising the project's integrity. Reallocate Resources: Consider redistributing resources (time, budget, personnel) to address stakeholder priorities while keeping the project viable. Agree on Compromises: Facilitate agreement on solutions that, while not perfect for all, represent a fair balance and advance the project.

Begin by comprehensively understanding the requirements and points of conflict among stakeholders Analyze the existing system, processes and inherent issues. Identify problems in the current system and understand business rules. Document discussions and observations from the system study meticulously. Finally, strive to reach a consensus among stakeholders through collaborative dialogue.

First of understand the conflict of interest by facilitating call among all stakeholders. Discuss the pros and cons of both the ideas and build consensus agreed and convinced by all stakeholders.

Clear communication and of course transparency are instrumental in preventing conflicts and keeping the project aligned with its objectives. Decisions should be proactively communicated that everyone is on the same page and understands their responsibilities.

Once a decision or solution is agreed upon, create an implementation plan that outlines the steps, responsibilities, and timelines. Communicate the decision to all stakeholders and ensure they understand the rationale behind it.

Keep in mind that managing conflicting interests requires ongoing communication and collaboration. Regularly assess the impact of decisions on stakeholders and be open to revisiting and adjusting the approach as needed. Additionally, document the decision-making process to maintain transparency and accountability.

If all the above doesn't work, remember the most useful design pattern: delegation. By delegating the decision to the Gods in the form of a trial by combat among the stakeholders, you are avoiding taking a decision and taking the best decision at the same time