Last News
The staff of Asclepius was for the Greeks the symbol of healing diseases. It is a rod with a coiled snake, which expresses the dual nature of the drug – medicine in the right dose and poison in excess.
For mankind, artificial intelligence can be a cure. It may lead us to situations we enjoy previews or scenarios where we suffer the consequences of its abuse.
The AI-rich future has us hiding incredible technological advances in massive amounts of data, which can create new jobs, accelerate the creation of vaccines, and give us tools against crime and climate change.
A less encouraging vision shows us massive unemployment, a low rate of re-employability of the people involved, and authoritarian governments that use this. technology to manipulate the opinion of its citizens. In addition, there is a greater economic disparity compared to countries that have the resources and talent necessary to train massive AI models.
LOOK: The engineer devises a magnetic bicycle that does not use gears or pinions | Video
The real danger of AI
So is AI medicine or poison? If we consult a doctor to understand the side effects of medicine, let’s pay attention to the scientists who are in charge of developing this technology.
For example i a recent interview with Le Point magazineYann LeCun, head of AI research at Meta, reflected on the danger to democracy represented by language models. “I don’t think it’s dangerous at the moment, but the reason could be that this is a closed system, so people can’t understand it, and researchers can’t investigate it. “. Yuval Noah Harari, author of Sapiens, mentions in the same conversation, “It can be great, but in the wrong hands it can destroy democracy. Because democracy is a conversation between people, and what if you have these systems that can have a conversation better than anyone.”
LOOK: What is the best soccer club in Peru? Read the ChatGPT answer
An even more urgent voice is that of Geoffrey Hinton. Far from being someone who seeks to attract attention with the predictions of the day, Hinton represents the calm voice of someone who was very popular three decades ago. mathematical mechanism (backpropagation) to It allows neural networks to learn based on errors. Hinton mentions to the Guardian that “We have discovered the secret of immortality. The bad news is that it is not for us.”emphasizing the limits of the human brain to transfer the information it learns. “Look at how things were five years ago and look at how they are now. Take the difference and project it into the future. That scares me”states
The words of these people represent the real danger of AI: the rapid pace at which it advances without apparent human control. But 5 years ago this technology surprised us with unprecedented accuracy in text classification and showed a impact on health with the commercialization of the first medical device that determines without human expression presence of diabetic retinopathy. Today, more sophisticated systems, such as GPT-4 and Midjourney, are already in question to replace their human counterparts in some tasks.
LOOK: Google will show automatic help messages for people looking for self-harm
When will we see these changes?
Interestingly, even the scientists who develop these algorithms have trouble predicting when AI will be at least as intelligent as humans. That happens we live surrounded by incremental processes, Therefore, it is difficult for us to understand reality in the presence of exponential changes.
In fact, the inventions that shaped earlier industrial revolutions, such as the steam engine, the combustion engine, and the personal computer, were limited by the physical limits of their own production process. That is, although the Internet has a profound effect on the global economy, the number of computers manufactured each year to access it is limited.
Instead, the amount of data we produce on the Internet, and end up feeding deep neural networks, follows an exponential behavior over time. Such a trend is also observed in the Data Age 2025 Reportdone by IDC.
“The main characteristic of the current development of AI is its models show more intelligent abilities when they get more data and computing power. Hence the versions of it GPT-2, GPT Chatand GPT-4 they seem to be much smarter in every version. That speed does not give us much time to understand the profound social implications of these algorithms and reduces the time for action to mitigate their potential risks. For this reason, no one in the scientific community could have predicted such a big change in 5 years, and today it is difficult to predict with certainty what will happen 5 years later.”.
LOOK: Google TV provides 800 channels for FREE: How can you access them?
However, we can find patterns based on previous experience. Like medicine, the massive use of this technology will go through several stages and affect different areas of society in different ways. At the start increase the productivity of many jobs and create new jobs in sectors where there is increasing demand and more data available to train AI systems. But it will also reduce jobs in those sectors where productivity increases, where we have data, and where the task is repetitive and constant. Scientists like Yann LeCun estimate precisely that this transition phase in which this new technology will lead to jobs can last between 10 and 20 years. In addition, as they try to create increasingly useful services, the data these models will maintain will be increasingly personal, starting with the web pages we visit, to later incorporate information on our activities physical and our geographical location.
Is AI a danger to Peru?
Before danger AI represents a tremendous opportunity cost. In other words, if we are facing the fourth industrial revolution, the cost of not generating an AI-based industry in the country is huge.
This transformation is driven by data, so it makes sense to start identifying the data that gives us a global competitive advantage. A government effort to collect and maintain non-confidential information generated in Peru’s main economic activities, such as mining, fishing, agriculture, construction and tourism. the first strategic step to create added value. The next step in creating scientific enterprises and projects that solve real problems is to give universities and the qualified public access to this information and computing credits. For example, it is not difficult to imagine a large number of tasks in agriculture and mining that are repeated daily in different parts of the country, and whose results can be predicted considering the details of the place to generate better budgets , access to providers, and reduce subjectivity in decision-making.
Bringing open data, code and computing to the population is perhaps our best bet to accelerate the development of intelligent systems against years of inaction to advance science and technology in the country. These changes will not come overnight and the government needs to organize them. through a regularly reviewed national AI strategy. An effort already started in 2021but which, remarkably, is still a decree law waiting to be approved.
History has shown us that it takes many years for governments to come to a consensus to address the negative consequences of each industrial revolution. From labor exploitation, environmental pollution, lack of data privacy, and potential large-scale automation. But it also showed us that, in the face of a common danger like the pandemic, society and its authorities can make quick decisions based on data and create policies that effectively benefit the entire population. In the midst of that uncertainty, what is real is change. We can watch from afar as a new industrial revolution unfolds before our eyes, or come together to try to be a part of it. We can import this medicine called artificial intelligence, or take concrete actions to create a new industry with a Peruvian brand.