Designing With Ethics: The Most Urgent Challenge in Artificial Intelligence

Artificial Intelligence does not create bias—it replicates it. That was the central warning from Manuel Pliego, Government Affairs Director at Microsoft Mexico, during his participation in Innovation Day. In light of the rapid advancement of these technologies, Manuel emphasized the importance of building systems that embed ethical principles from their very design.

According to the OECD Declaration on AI (2019), automated systems must be fair, transparent, and safe—especially when they affect fundamental rights such as freedom, privacy, or equal opportunity. When these conditions are not met, algorithms do not correct inequalities: they reinforce them.

AI as a reflection of our society

Manuel explained that bias already exists in the data we use. “Are biases invented with AI? Are they created by AI? No. Biases already exist in society. What happens when AI is poorly designed? It replicates them.”

He illustrated this phenomenon with an example in the United States, where an automated system granted early release to inmates. Since it was fed with historical crime statistics, the result was a prioritization of white inmates’ release, ignoring the racial bias embedded in the data.

“It’s a mirror, like Black Mirror, the Netflix series. A mirror that shows us what is already there and that we must face,” he said.

He also warned of the risk of operating with algorithms that lack traceability. “If you don’t know how decisions are being made, we have a problem. If you can’t explain it, we have a problem.”

Microsoft and the path toward responsible AI

Manuel shared that Microsoft has taken a long journey toward building ethical AI, based on principles such as fairness, inclusion, privacy, transparency, and accountability. Since 2016, the company has promoted multiple initiatives, including the creation of the Ether Committee, the Responsible AI Office, and standards such as the Responsible AI Standard v2.

This approach not only covers internal governance but also active participation in multilateral agreements such as the Bletchley Declaration (2023) and the United Nations General Assembly resolution on artificial intelligence (2024).

The urgent need to regulate AI misuse

During his talk, Manuel highlighted the most pressing ethical dilemmas. Among them, he pointed to real-time mass surveillance through state-run facial recognition systems. “All that creates is the Big Brother of George Orwell’s novel 1984… cutting off your democratic freedom.”

He also referred to automated decisions that prioritize questionable criteria. He recalled the case in a film where a robot saves an adult instead of a child. “What did it take into account? What if it considered things like ‘he was a man,’ or worse, ‘he was economically productive.’ That’s when we are no longer acting correctly as a society.”