Resources

We interact with AI much more than we sometimes realize: from chatbots like ChatGPT to cleaning robots like Roomba, voice assistants like Alexa or Siri… It makes tasks easier and provides information in a simple and fast way, but it also comes with risks and challenges.

Artificial intelligence (AI) has been integrated into our daily lives for a long time. When we think of AI, the first thing that usually comes to mind is generative technology like ChatGPT, but in reality, we interact with it more than we sometimes realize through GPS devices, cleaning robots like Roomba, voice assistants like Alexa or Siri… It makes tasks easier and provides valuable information in a simple and fast way. However, it also comes with risks and challenges

Artificial intelligence simplifies and facilitates complex or cumbersome processes and, ultimately, has the potential to improve our lives. “It can perform calculations faster, make decisions more quickly, see where we humans cannot, and analyze documents much more rapidly,” explains Miquel Àngel Prats, professor of Educational Technology at Universitat Ramon Llull. “Many of us have lived without the internet and have witnessed how its arrival profoundly changed the way we interact, communicate, and access information. The phenomenon of artificial intelligence is precisely that kind of revolution,” he states.

However, like any rapid and profound change, artificial intelligence also raises challenges and uncertainties, and its uncontrolled use can even have a negative impact on our societies. It is also important to remember that behind this technological leap, there is an industry with its own interests that may conflict with the common good. “A market has developed at a dizzying pace and without control,” warns Karina Gibert, an expert in artificial intelligence and computational statistics at the Polytechnic University of Catalonia. “Many so-called intelligent tools have emerged that are actually just smoke and mirrors—they don’t contain real AI, just advanced statistics and heavy marketing. Others are indeed AI but have not been properly trained or validated, leading to malfunctions or biases that have caused major setbacks and negatively impacted certain sectors of the population. Some have even been trained with illicit data that the rightful owner never consented to share. All of this paints a picture of a somewhat reckless landscape that carries significant dangers, as we may be violating fundamental human rights,” Gibert states.

Be careful not to confuse generative AI (ChatGPT, Gemini, Copilot, Perplexity…) with a search engine like Google or Yahoo. Their function and operation are not the same, even though both provide information. The former organizes existing information, while the latter creates new information based on prior data. “Generative AI responds with sentences it constructs using probabilistic criteria based on everything it has read. In other words, it predicts which word fits best after a partially formed sentence and builds the text accordingly. Generally, it sounds good, as these systems tend to produce very convincing and well-structured texts. However, we must remember that they do not understand anything we ask them, nor do they truly grasp the meaning of their responses,” warns Gibert. In fact, generative AI focuses so much on making texts sound ‘impressive’ that “by nature, when it doesn’t know a concept, it simply makes one up,” Prats states.

It is important to consider that the foundation of generative AI’s knowledge is its training with massive amounts of data—any type of data, even incorrect ones. This is because these systems are designed to understand human language. This training process means that they also learn a lot about the world, but the information they provide may contain incorrect dates, inaccurate data, or misunderstandings. These are known as ‘hallucinations,’ errors caused by insufficient training data, incorrect assumptions, or biases related to race, gender, or culture. Some people say, “data speaks for itself,” but data only speaks for itself if it is collected correctly, and even then, it can say whatever you want depending on how you interpret it. “It’s the paradigm of prediction,” says Albert Sabater, Director of the Chair – Observatory of Ethics in Artificial Intelligence of Catalonia (OEIAC) and professor at the University of Girona.

As we mentioned, AI is trained using massive volumes of data, including, in many cases, personal data whose use has not been authorized by its owners. It wouldn’t be the first time that internet users unknowingly give up their data without realizing the commercial use that is later made of it. However, in many cases, the companies behind these technologies have not been sufficiently transparent in disclosing whether the datasets used for training include private user information without their consent.

On the other hand, we should also be aware that, for our security and privacy, there is certain information we should never share with AI. We should not provide data that could identify a specific person or disclose personal information such as full name, date of birth, address, ID number, phone number, or email. Of course, we should also avoid sharing banking details, precise vacation dates or travel itineraries. Additionally, we should never provide passwords or ask AI to generate them, nor should we share medical records or protected information. At a business level, it is recommended not to disclose programming codes, product information, trade secrets, or other sensitive data to AI.

Every time we ask AI a question, we increase our carbon footprint, so the future of the planet also depends on how responsibly we use this technology. And although new generative AIs, such as the Chinese Deep Seek, claim to be more efficient and therefore produce fewer greenhouse gases, we must be aware that our choices have an impact on the climate crisis. “Generative AI requires a lot of computing power, and with more computing, you need more data centers, which consume a significant amount of energy and also use a considerable amount of water,” warns Sabater.

“These tools are so attractive, so seductive, that they almost trap you,” admits Prats. The speed at which they provide acceptable results, their ease of use, and the many tasks they can perform may lead us to the almost addictive temptation of leaving our creativity in AI’s hands. Why should I create something original and unique if AI can generate a correct version in just a few seconds? Why should I put in the effort? “We need to self-regulate,” argues Prats. “It’s very easy to get hooked on the screen and constantly turn to AI. That’s why we need to have a well-structured mindset and set an example for children and young people.”

Without falling into alarmism, it must be acknowledged that AI provides unprecedented ease for plagiarism, and its widespread use is loosening moral (and also economic) conventions—let’s not forget that many professional sectors are dedicated to text creation. “Why have people been told that they need a generative AI model to produce texts they haven’t written, claim them as their own, and put their name on them? That is plagiarism. Why should I tell my child that this is acceptable when I know it’s not? We have reached a point of widespread corruption,” laments Sabater, who warns that this malpractice is increasingly spreading into the world of scientific research. “For many years, the demands placed on academics have been completely unsustainable. And some are using tools like ChatGPT and others to be much more productive,” Sabater admits.

According to the experts consulted, many companies have generated exaggerated expectations about their products for commercial reasons. “People sometimes get caught up in FOMO, in factors that are less about technology and more about the economic side of things,” says Sabater. AI is not suitable for everything, nor should we use it in all cases. As the saying goes, you don’t use a cannon to kill a fly. “The problem lies in the deployment of a methodology that is sometimes not the best solution for a given problem,” Sabater explains.

As we have seen, AI has the ability to generate credible but not always truthful information, something that is creating a “disturbed and polluted digital ecosystem,” in Sabater’s words, where almost anyone can spread misinformation. “We are clearly giving up part of our ability to produce quality information to companies whose only goal is to generate revenue. So far, they don’t seem very interested in the contamination their tools are creating in the digital ecosystem,” the expert laments.

No, technology won’t steal your job, but business decisions might. A study by Nexford University analyzing how artificial intelligence will transform the job market between 2024 and 2030 highlights that certain jobs are at higher risk of automation. Administrative work, customer service, data entry, and accounting are among the most affected, as AI can perform these tasks more efficiently and quickly. Creative jobs such as designers, advertisers, copywriters, translators, actors, and singers are also impacted. “We see in some companies’ agendas the intention to market this generative technology as a replacement for certain professions,” warns Sabater.

In fact, some professions have already begun to mobilize against being ‘replaced’ by AI. In September 2023, the Writers Guild of America ended a 148-day strike after reaching an agreement that, among other things, established that writers could only use generative AI as a tool, but a work could never be created or signed entirely by AI or have this technology as its primary source. In the United Kingdom, on February 25, more than 1,000 British musicians supported a silent album as a form of protest against a new law that makes it easier for major tech giants to bypass intellectual property rights and use their compositions to train new generative AI models.

“We still need a lot of awareness about this phenomenon, so we must demystify it and approach it naturally. That’s one aspect. On the other hand, we need to emphasize security and privacy precautions,” says Prats. In this regard, education is essential. First, in the classroom, where teachers, according to Prats, must understand the current context and be able to assess the potential, risks, and limitations of AI use. But families also play a crucial role. “We need to work with families so they stay informed and can help their children develop critical thinking. It’s not about banning AI but finding a balance,” he adds.

On the other hand, Karina Gibert emphasizes the importance of transparency in AI usage. We have the right to know when we are consuming a product generated by artificial intelligence. “The idea is to make AI usage transparent so that everyone knows when AI is involved in a process, can request explanations on how it reached a certain outcome, and can appeal in defense of their own rights,” the expert states.

In the end, AI is here to stay, and we must learn to coexist with it and empower ourselves to make it work for us, not the other way around. “We only have two options,” Prats argues. “Either we become Mennonites and retreat to the mountains, effectively disappearing from the world, or we acknowledge that we live in a highly digitalized world and empower ourselves to understand when and how technology should be used.”