1.What is the new study reveals?
According to a new study, there is a linguistic bias in English that causes us to “improve” things by adding to them rather than by subtracting from them. This bias is so pervasive that AI chatbots have been trained to use it. The word “improvement” is more frequently used in conjunction with addition than with subtraction. This may cause us to make decisions that hinder efforts to improve things.
2.Which kind of bias the scientists saw in a language?
Their research relies on previous research that has demonstrated that when people want to improve themselves, they often add items. They discovered that the bias is strongly ingrained in the English language. For example, the word ‘improve’ is more similar to the words ‘add’ and ‘increase’ than it is to the words’subtract’ and ‘reduce’. The study also discovered that other verbs of change, such as ‘to change,’ ‘to modify,’ ‘to revise,’ or ‘to enhance,’ behave similarly, and if this linguistic addition bias is not addressed, it can make things worse rather than better.
3.How it is ingrained in AI Chatbots?
The bias is so engrained that it is embedded into AI chatbots. The researchers asked GPT-3, ChatGPT’s predecessor, what it thought of the term “add.” “The word ‘add’ is a positive word,” it responded. Adding something to something else usually improves it. For example, adding sugar to your coffee will most likely improve its flavour. You will most likely be happier if you make a new friend.”
We should all be mindful of the positive addition bias in the English language. It can impact our actions and make us more inclined to add additional layers, levels, and objects when we might benefit from deleting or simplifying. Perhaps the next time we are asked at work or in life to provide suggestions for changes, we might pause for a moment to examine our options.