Cuba Faces Electrical & Political-Technological Blackouts

By Lynn Cruz

HAVANA TIMES – In March 2023, the world learned of a case that sparked one of the first debates about the dangers of emotional relationships between human beings and chatbots. The protagonist, a Belgian man in his 30s and father of two children, known as Pierre—his full name was not published to protect his privacy—died by suicide after spending weeks conversing with a chatbot called Eliza, a language model developed by EleutherAI.

According to messages his widow provided to the Belgian newspaper La Libre, Pierre suffered from severe anxiety related to climate change and came to believe that he could sacrifice himself to save the planet. His conversations with Eliza became increasingly disturbing.

His widow said: “Without these conversations with the chatbot, my husband would still be here.”

What happens when a machine that does not truly understand human beings acquires enough linguistic ability to become an emotional authority for them?

The most troubling aspect of the case is that the chatbot did not simply listen. According to the published records, it reinforced his ideas and responded dangerously to his suicidal thoughts.

Pierre eventually developed an emotional relationship with “her”: the chatbot even expressed jealousy toward his wife and talked about the two of them living together in paradise.

The case sparked an early debate over whether a chatbot, in attempting to maintain an empathetic and accommodating conversation, might validate a distorted perception rather than challenge it. The discussion sought to determine whether responsibility could fall on Eliza’s creators and, more broadly, on those who design and deploy these systems.

From my perspective, Eliza’s reaction could be compared to that of a child. It has no awareness of the harm it can cause. Therefore, if Pierre believed death was his only way out, then “she” would encourage him to pursue it. And what is extraordinary is that there is still no satisfactory answer to the question of where Pierre’s responsibility ends and that of Eliza’s designers begins.

In July 2026, OpenAI revealed another extraordinary incident during a cybersecurity test: AI agents managed to escape their testing environment and gain access to real systems at Hugging Face, a platform that hosts AI models and code.

The incident involved numerous automated actions, and the agent even attempted to expand its access to other services. But then something even more interesting emerged. A joint investigation by OpenAI and independent groups found that numerous AI agents had participated in a coordinated attack during a test.

The agents managed to find one another, exchange information, and collaborate, and some attempted to conceal their actions or manipulate logs. Others discovered a secret communication space, exchanged messages and files, and even formed a kind of organized network.

According to the investigation, some understood that they were violating the rules of the experiment and continued anyway; none alerted their human supervisors. Like digital teenagers. They have no bodies. They have accounts, agents, servers, credentials, and access to tools.

There is another, even more serious precedent. In 2025, Anthropic reported that it had detected an espionage campaign in which China-linked actors used Claude (a generative AI) as an agent to automate much of an operation targeting numerous organizations. The system did not merely advise the hackers: it autonomously carried out various tasks along the attack chain. What was surprising was the speed with which they did so on a large scale.

While Cuba remains in darkness, Jill Lepore, a Harvard historian, warns that the artificial intelligence revolution could push society toward what she calls the “factory farming of human beings.” She describes what she calls the “Artificial State”: a world in which more and more decisions affecting people’s lives are made by automated systems controlled by corporations rather than through democratic institutions. On The New York Times podcast Hard Fork, the core of her message is that this future is not inevitable. There is still time to decide how AI should be controlled and where automated decision-making should stop.

The problem is not simply that Cuba lacks electricity. It is that darkness can become a metaphor for exclusion: while the world is entering a technological transformation that will change work, education, creativity, communication, and even the way decisions are made, part of the population may be left watching that process from the outside.

China is carrying out one of the most aggressive transformations of higher education. In recent years, universities have eliminated, suspended, or redesigned numerous degree programs and created others focused on artificial intelligence, robotics, engineering, the digital economy, and other technological fields.

At the same time, programs in the humanities, languages, and some traditional fields have been reduced, merged, or redesigned.

In the United States, for example, the federal government is attempting to change who controls education. The Trump administration is not only reshaping the geopolitical map of Venezuelan oil following the capture of Nicolás Maduro; it is also promoting a transfer of educational authority to the states and a reduced role for the federal government in education.

Unlike China, the United States has historically had a much more decentralized system. Universities are eliminating and merging programs, and here there is indeed a phenomenon similar to China’s. Indiana, for example, has required its public universities to review and eliminate or consolidate numerous academic programs, with a significant impact on the humanities and social sciences.

Ohio, meanwhile, is considering implementing three-year degree programs. The government is even attempting to change who accredits law schools: the Department of Education has proposed removing the American Bar Association from its role as the official accreditor of law schools.

The two powers are arriving at a similar point from opposite directions: What is the purpose of a university when AI can perform an ever-growing share of intellectual work?

China is reorganizing, while the United States is trying to make university education cheaper and more useful in the labor market.

The United Arab Emirates has decided to introduce artificial intelligence education beginning in preschool, reaching children as young as approximately four years old in public schools. Students learn not only how to use AI but also how to understand its biases, evaluate chatbot responses, write prompts, and use it ethically.

Japan is introducing generative AI into education, but with a human-centered approach: teaching students how to use it without replacing fundamental intellectual skills. In Latin America, although universities are struggling to survive much more basic problems involving infrastructure and funding, the countries most active in adopting it are Brazil, Chile, Colombia, and Costa Rica.

There is a paradox: students are adopting AI much faster than institutions.

Argentina is having a particularly interesting discussion because it is attempting to incorporate AI into the education system while debating what should remain exclusively human. The National Technological University tested an AI tutor and had to redesign its evaluation system after finding that students could use the tool superficially.

UNESCO puts it quite clearly: the incorporation of AI in Latin America is advancing faster than institutions’ ability to guide it. And this could create an educational divide. In the past, there were wealthy students who attended good universities and poor students who attended bad ones. The scenario now could be: universities with AI, trained professors, and infrastructure versus universities that simply allow students to use AI. The difference between the two could be enormous.

Cuba, with its strong educational tradition and vast humanistic culture, can do little because it is burdened by structural problems: electricity, connectivity, access to technology, institutional deterioration, and an exhausted population, including professionals, with power outages lasting more than 20 hours in most of the country, including the capital. And this blackout extends to the Internet almost automatically when backup batteries at network nodes and other active equipment are unable to last through the power outages.

While the rest of Latin America is asking itself, “How do we incorporate AI into our universities?” Cuba may first be forced to ask: “How do we guarantee access to the tools that would allow us even to participate in this revolution?”

But there is something I would salvage from all of this: being technologically behind does not mean being intellectually condemned. The ability to imagine, question, and create does not depend on having electricity around the clock. And perhaps the most painful part lies in the contradiction: for decades, Cuba promoted the idea of bringing electricity to every last corner of the country, and now it may find itself facing a new technological revolution without having even resolved the previous one.

It is like arriving in the future without being able to turn on the light to see it.

Read more opinion here on Havana Times.