Guarding against Bias in ChatGPT and Travel AI Tools for Fair and Accurate Results

Concerns about the need to regulate artificial intelligence (AI) technology, developed by companies like OpenAI, Google, and Microsoft, have prompted discussions about the potential harm of unintended bias in AI models. The director of data science for Placemakr, Jyotika Singh, believes that biases in AI models can arise from assumptions used in developing models or the data used to build and train them. One example of bias in data involves ChatGPT, which has been built using data taken from the internet, including social media and books, that contains a lot of bias. Singh suggests that the data’s bias inevitably influences the ChatGPT model. Singh also noted that building diverse AI teams that work on developing AI models is crucial for guarding against the impact of bias. Finally, to mitigate the negative impact of biases, companies often open their products to a limited set of individuals and collect feedback, allowing them to identify and promptly correct issues.