At EdukIΔ, we believe that artificial intelligence (AI) can become a powerful ally for researchers: not to replace human curiosity, but to amplify it. In this post, we explore how AI is being applied in scientific research today, what opportunities it offers, particularly in Latin America, and what ethical challenges we must address.

🤖 What does AI contribute to research?

AI provides tools for:

• Processing large-scale information: Imagine reading, filtering, and organizing hundreds or thousands of articles in minutes.
• Discovering subtle patterns: It detects correlations that a human might miss.
• Generate hypotheses and suggest methods: Some tools can suggest new lines of research and experimental strategies.
• Assist in writing and dissemination: from summarizing findings to preparing more accessible "general audience" versions.

The key: AI offers intellectual and operational support, but human judgment remains essential in every decision.

🌎 In Latin America: challenges and possibilities

The Latin American panorama has particular features that should be recognized.

• According to the Latin American AI Index (ILIA 2025), Brazil concentrates more than 90% of the region's computing capacity (indexelatam.cl).
• In the education sector, digital inclusion remains a barrier: in several countries, one in ten students does not have access to a computer or stable connection at their institution (blogs.iadb.org).
• Even so, there are emerging initiatives: in Latin America at least 26 projects applying AI to the educational field have been identified, although many lack local adaptation (profuturo.education).
• Recent research using AI in regional education shows interesting results. For example, a study on school efficiency gaps applied hybrid techniques of “machine learning explainable + DEA analysis” to understand how private and public schools use their resources (arXiv).
• Another study explored the determinants of academic resilience in Latin American students, using interpretable methods (such as SHAP). This identified variables such as devices in the home, teacher quality, and digital access as key (arXiv).

These studies show that in Latin America, AI applied to education and research is not just an aspirational goal, but is already present and producing results. insights valuable.

⚠️ Ethical challenges and limits to consider

Although AI has great potential, it also presents risks:

1. Biases and unbalanced dataIf the data used to train models is biased (by geographical areas, languages, social classes), the predictions may reinforce inequalities.
2. Lack of transparencyIt is not enough for it to "work" — it is necessary that others can reproduce and understand how it arrived at a result.
3. Excessive dependence: Blindly trusting what AI suggests can diminish human creativity or criticism.
4. Unclear publishing guidelinesIn many Latin American journals, it is not defined whether the use of AI in the writing or generation of content should be declared.
5. Resource and training gapMany institutions lack the computing infrastructure or trained personnel to use AI rigorously.

UNESCO, for example, promotes human-centered AI for education, aiming for equity and inclusion in the face of technological inequalities (UNESCO).

🚀 Recommendations for researchers (and for EdukIΔ)

To take advantage of AI without losing control, I suggest:

• Experiment with small prototypes before scaling up: test simple models first.
• He prefers tools from open source y institutional repositories, preferably open access where possible, for greater local control and adaptation (which also enhances technological sovereignty).
• Document everything: what data you used, how you processed it, what parameters you configured. This makes reproducibility easier.
• Whenever AI is involved in results or writing, it declares its role and limits its use to support, not main authorship.
• Promotes internal training: workshops, seminars and peer sessions to build shared criteria.