As organizations bring AI into real project environments, new ethical and organizational challenges appear—such as unclear ownership of decisions, hidden bias in AI outputs, over-reliance on automation, and the risk of making inaccurate or unfair project decisions. How should we manage these risks and define accountability?
Saving Changes...
Sort By:
Luis BrancoCEO| Business Insight, Consultores de Gestão, LdªCarcavelos, Lisboa, Portugal
AI errors don’t create an accountability gap, they expose one. When an AI output is wrong, the failure is never the machine; it is the human system around it. Project leaders remain accountable for judgment. Organizations remain accountable for governance, data quality, validation, and ethical boundaries. AI plays a role, but never carries responsibility, because responsibility requires intention, context, and conscience. The real risk is not “AI making mistakes.” It is humans over-trusting automation, skipping verification, or deploying AI without clear decision rights, escalation paths, and ethical safeguards. If we want AI to enhance project decisions rather than distort them, we need three things: clarity of ownership, disciplined human oversight, and a governance culture where learning and verification are non-negotiable. Accountability does not disappear in the age of AI. It becomes a leadership discipline. Saving Changes...
AI can support decisions but it cannot carry accountability. When errors happen, the responsibility stays with the people who designed, validated, and used the tool. Clear governance, human review steps, and transparent data checks keep AI in its proper place. The team owns the outcome, and AI remains only an assistant. Saving Changes...
PMO Leader | Speaker & Mentor | Content Leader – PMOGA Latin America
Hub| Catholic University of UruguayMontevideo, Montevideo, Uruguay
Responsibility for AI errors cannot fall on the technology itself, but rather on the people and organizations that design, implement, and oversee it. To manage risks, it is essential to establish clear governance structures, define accountability roles, and maintain human oversight in critical decisions. Data transparency, bias detection, and ethical training of teams are essential to ensure that AI adds value without compromising fairness and trust.
Program Manager| HARPER SRLSanto Domingo / Distrito Nacional, Dominican Republic
Well, AI can assist in decisions, but it cannot own themm so accountability stays with humans, those who design, configure, validate, and approve the outputs. The real risk isn’t AI making an error; it’s teams treating AI as infallible. Clear governance, human review, and transparent decision chains are what keep accountability where it belongs: with the organization, not the algorithm. Saving Changes...
Amit KhedekarConstruction & Engineering Specialist| Fluor Daniel India Pvt. Ltd.Mh, India
Truly, as said by all, AI is a creation of humans. It behaves, responds and delivers as fed. So, AI totally depends on what is put inside it. Therefore, the three traits responsibility, accountability and even reliability cannot be questioned on AI. It's the one who designed and approved it and has to own the above traits whatever the results maybe.
I wasn't going to reply, as there are already multiple good responses that answer the question. I'd like to take a step back and respond to a broader question. There's a new tool. Who is responsible for errors made when using the new tool? I see two possible answers to this question: 1) the person who created the tool, and 2) the person who used the tool. However, it is important to distinguish between who is responsible for fixing the tool so that it produces fewer errors and who is responsible for the damage caused by using the tool. In some cases, the developer of the tool should be accountable for the damage. If a tool is faulty and injures the person using it, and the person is using it as instructed, the toolmaker is usually held accountable and either fixes future versions of the tool or ceases production. If there is an expectation that they end user should verify the output of the tool before using it, and the user does not perform this step, the end user is accountable for not performing the validation and the resulting damage. There are multiple examples of tool usage and resulting problems. Here are a few.
The London Whale incident at JP Morgan - was unchecked risk taking and billions in losses MS Excel's fault?
Was the crash of Air France 447 the autopilot's fault?
Is it the GPS' fault when a driver follows the instructions down a boat ramp and into a lake?
It seems like every few years we need to relearn the dangers of over-reliance on tools. Life is not something we can automate, however nice it MIGHT be to go through our days on autopilot and not be accountable for the choices we do and don't make. Approaching things like this is just a disaster waiting to happen.
Sergio Luis ConteHelping to create solutions for everyone| Worldwide based OrganizationsBuenos Aires, Argentina
First of all AI is a board term. Second, AI does not make errors. AI outputs are always probabilistic and the final decision always rest on human beings. If you are talking about generative AI then Responsible AI component is a must. Saving Changes...
Community Champion
