I can already no longer imagine a society without AI. At an incredible pace, AI-based systems are being integrated everywhere to make our world more efficient and faster. On a personal level, more and more people are embracing AI tools to simplify their daily lives. Like 180 million others worldwide, I use the AI tool ChatGPT and am continually impressed by its ability to analyze and reason with an efficiency far beyond my own. I often turn to it for brainstorming sessions while studying, creating schedules, or even finding inspiration for dinner. It feels like an inexhaustible source of knowledge.
AI: Not a Magical Solution
However, AI is not an endless source of magical intelligence. It is the result of the vast collection of digital data that we, as a society, have created over the past decades. Sometimes, it seems as though we forget that AI is not as objective as it appears. Everything an AI system knows comes from humans. This is why I occasionally question the rapid pace at which AI is becoming a fundamental part of our society. Should we even want this? As AI continues to develop, it remains inextricably linked to who we are as humans.
Governments and AI: A Balancing Act
Governments face the challenge of managing growing populations efficiently. It is, therefore, understandable that AI is becoming an increasingly important part of organizing our societal systems. Yet, I wonder whether we should embrace the use of AI with such enthusiasm or if we should instead put the brakes on its ubiquitous implementation.
In recent years, several examples of failing AI systems have emerged, often with severe consequences. Take, for instance, the childcare benefits scandal in the Netherlands. This shocking event led to the collapse of the Dutch government in 2021. The problem? An AI system used to identify tax fraudsters was found to operate in a discriminatory manner. Tens of thousands of citizens were wrongly labeled as fraudsters, partly because the system based its assessments on ethnic background and nationality.
This had enormous personal consequences: families fell into financial hardship and had to repay large sums of benefits, even though they had done nothing wrong. It was a painful example of how AI can reinforce existing prejudices rather than solve them.
Another notorious example is the use of predictive policing technologies, such as PredPol in the United States. This software predicts crime based on historical arrest data. The problem is that the data in these datasets are collected by humans and often already contain biases. This results in disproportionate police presence in low socioeconomic neighborhoods, where minorities often live. This perpetuates a vicious cycle of inequality and systemic racism.
These examples show that AI is not as objective as we sometimes think. Even a tool like ChatGPT, which I use extensively, is not free from biases. For instance, ChatGPT has been found to lean toward a left-liberal bias in political statements. While this may seem harmless at first glance, it illustrates that AI systems are not neutral. They are based on the data and beliefs of the people who developed them.
AI and Everyday Use: Convenience with Caution
More and more people are using AI tools not just to summarize texts or find recipes but also as a source of information, a brainstorming partner, and even a therapist. While this can be incredibly convenient, we must remain critical. An AI system is not an all-knowing, objective advisor but a reflection of the world we have created. All our flaws and biases are part of it.
The future of AI is, in many ways, the future of humanity. It is up to us to ensure that it aligns with the values of fairness, inclusivity, and accountability. If we succeed, AI can be a powerful ally in building a better world. If we fail, it may only deepen the divisions we already face.
The question, then, is not whether we can imagine a world without AI—but whether we can imagine a world where AI truly serves everyone. That vision demands more than innovation; it demands responsibility.
Sources:
https://www.universiteitleiden.nl/nieuws/2023/03/chatgpt-heeft-linkse-voorkeur-bij-stemwijzer
Recent Comments