There is no doubt that artificial intelligence (AI) has made our everyday work and the way we gather information much easier. However, new studies warn that the information analyzed by artificial intelligence and the details it summarizes for us have the potential to reshape the human brain’s memory system itself. This is not merely a matter of technological advancement. It is a major challenge to our thinking ability and our capacity to think independently.
In today’s fast-paced globalized world, people do not have time to read long articles or multi-page research reports in full. As a result, we ask AI tools like ChatGPT and Google Gemini to “summarize this briefly.” They, too, condense the information in an instant.
Researchers point out that when artificial intelligence tools summarize information, they make two major mistakes:
1. Leaving out important information:
AI tools leave out the most important details of an article. A study using various prompts and models found that about 51.6% of critical information is not included in AI-generated summaries.
2. Generating false information:
They present something that does not exist as if it did, or describe it inaccurately in summary form. Because of their trust in technology, people also believe that what AI provides is true.
The human brain is very fascinating, and AI can erode human memory. At the same time, it is also easily deceived. Even if we have directly witnessed an event, our brains become confused when we later read AI-generated summaries containing incorrect information about that event. As a result, people may forget even the real things they directly saw and knew after reading information provided by AI.
Study and findings
The results of a study conducted among 331 people by researchers from Georgetown University and the University of Washington in the United States confirm this.
Participants in that study were shown an animated video of a car accident at an intersection. After 24 to 48 hours, they were shown a summary note related to the AI-generated video. Some were given an accurate summary, while others were given a summary containing incorrect details. They were then asked to recall the event and say who was at fault in the accident. The study’s findings:
> 83.6% of those who read the accurate summary correctly recalled the actual event.
> However, only 44.8% of those who read the AI summary containing incorrect details remembered the truth accurately.
Even when participants were given the warning in advance, “This is an AI-generated summary, and it may contain errors,” the misleading effect did not diminish.
Although we know that AI can make mistakes, our unconscious mind records the information provided by AI as if it were true memory. Our immense trust in technology is also a major reason for this.
Researchers’ warning :
According to Matthew Sim, an assistant professor at Georgetown University who led this research, “It is often expected that humans will step in and correct the mistakes of AI technology. However, our study suggests that human memory itself may be distorted by the mistakes made by AI instead.”
Which fields are at risk?
Court - Criminal:
If a witness who saw a crime is shown an AI-generated summarized report, they may lose confidence in their own memory and give false testimony.
Medical field:
A small mistake when AI summarizes a patient’s medical report can lead to incorrect treatment.
News - Politics:
AI-generated summaries and news circulating on social media can make people believe false information and fake news are true.
Artificial intelligence tools are only assistants meant to help us. They cannot replace our brains. Accuracy is more important than speed. Therefore, when making important decisions or learning about complex subjects, it is safer not to rely only on AI-generated summaries, but to read the original sources ourselves directly. AI may save us time. But we should not allow it to determine our memory.