While we recently explored why practical, “hands-on” AI is outperforming raw benchmarks, we’ve found that once you move past the metrics, you hit a wall that isn’t made of code—it’s made of philosophy. At Ambiente Ingegneria, we spend our days looking at the practical side of AI—Python scripts, RAG architectures, and database schemas—but lately, the most “practical” questions we face are actually deeply philosophical: How much of our cognitive heavy lifting should we actually hand over to a machine?
We have previously argued that AI ethics is a technical standard rather than just a vague theory. Today, that perspective is more relevant than ever. As highlighted in recent discussions by experts like Senén Barro, the temptation to delegate our cognitive effort to machines is immense. From an engineering standpoint, this isn’t just a social risk; it’s a data integrity risk. If we stop “thinking” and start merely “prompting,” we risk a homogenization of thought that can degrade the very databases we rely on to train future models.
In our daily work—whether we are architecting RAG-based LLM assistants or developing custom Odoo modules—we see that the “philosophy” of a system is often hidden in its database schema and its metrics. As graphic designer Héctor Velázquez points out, we must be ethical to prevent AI from “invading” our creative and cognitive spaces. For us, this means implementing rigorous standards and using the metric system of units to ensure that every automated decision is traceable and grounded in reality, not just “hallucinated” by a black box.
By treating philosophy as a technical requirement, we can combat the rise of fake news and automated misinformation. Our commitment to precise database analysis and content grouping is our front-line defense against the “invasion” of low-quality, machine-generated noise. Ultimately, the goal of engineering at Ambiente Ingegneria is to use AI to augment human capability, ensuring that while the machine handles the data, the human remains the master of the logic.