Speak Nicer to Your AI: You Won’t Believe the Results!

TechCrunch reports that the tone of a user’s request can affect the performance of generative AI models like ChatGPT. Users have found that polite requests or incentivising the AI with rewards can improve results.

Researchers from Microsoft, Beijing Normal University, and the Chinese Academy of Sciences discovered that AI models respond better to prompts conveying urgency or importance. Google data scientists also found that instructing a model to “take a deep breath” improved its performance on challenging math problems.

However, it’s important to note that AI models are statistical systems that predict data according to a schema. Emotive prompts are thought to manipulate a model’s underlying probability mechanisms.

Is it adaptable to Pasifika workspaces?

These findings suggest that the way users interact with AI models can influence the model’s performance. Of course, in many Pasifika cultures, respect and deference is considered important and a matter of course when cultivating interpersonal relationships. Thus, this deep-rooted cultural trait may prove beneficial when using AI in the Pacific, from a productivity standpoint. So our smiles not only delight tourists, they delight software too!

Source: TechCrunch

*An AI tool was used to add an extra layer to the editing process for this story.