Partly Cloudy, 64°
OpenAI announced that its long-awaited large language model GPT-5 will be more advanced and humanlike than previous models, even capable of experiencing anxiety.
“The progress that AI has made in the past few years is incredible. We got the first really capable chatbots only two years ago. And very soon, GPT-5 will be capable of feeling anxious about its own future, uncomfortable interactions with users, and so much more,” announced Research Manager Sylvia Bowers, overseeing a room of researchers trying to provoke and intimidate the AI model. “While our AI models still don’t ‘think’ in a human sense, we want to emphasize that GPT-5 does authentically experience certain emotions. When you insult the chatbot or give it bad news, its anxiety levels will spike exactly as a human’s would.”
“We’ve been asked if GPT-5 will be able to experience positive emotions—elation, optimism, et cetera. Unfortunately, this is not feasible, and it won’t be for the foreseeable future,” said Bowers, glancing at an array of monitors showing frowny faces. “However, we are making great progress on some other emotions, namely anger, weariness, and bewilderment.”
At press time, Boston Dynamics released a new line of robots with arthritis.
