AI Counters False Beliefs

People are fond of conspiracy theories, e.g. Covid pandemic was a deliberate attempt to control population or 9/11 attempt was executed by the US itself. Artificial intelligence (AI) can dissuade people of their beliefs in conspiracy theories.

There was an experiment to prove that strongly held beliefs can also change when convincing evidence is presented. Some 2000 volunteers asked ChatGPT4 Turbo to test their conspiracy beliefs. Volunteers were to rate their beliefs on a scale of 0 per cent to 100 per cent. They asked volunteers to substantiate their beliefs. The researchers asked the LLM to persuade people to reconsider their beliefs. Surprisingly, this worked. There was a drop of 20 per cent on an average of their faith in false beliefs. Some 25 per cent volunteers dropped their belief level to less than 50 per cent. The results were an eye-opener. The main reason to detach from their beliefs was the availability of more information, thus diluting their steadfastness in the veracity of their beliefs. It was a competition between evidence and counterevidence. The results have been published in Science in September 2024.

There are beliefs about alien life on the earth, the sighting of UFDs and assassination of political leaders. There are reporters who present their guess work as facts. There are rumours of people eating their pets.

Logic and evidence are powerful enough to dissuade people from holding such beliefs.

print

Leave a Reply

Your email address will not be published. Required fields are marked *