ChatGPT, a chatbot developed by the American company OpenAI, has re-sparked a debate on the risks of artificial intelligence (AI) technologies.
Chatbots like ChatGPT can create an impression of having thoughts and trick people into thinking they give credible information. The public has voiced concerns that the AI technology behind chatbots can be misused to deceive people. Irresponsible parties can use the tech to spread false information and make believable scams.
Some startups that develop chatbots provide disclaimers about the byproducts of their tech. OpenAI, for example, warns users about incorrect information, biased content and negative instructions that may be generated by ChatGPT.
ChatGPT is an AI-based chatbot that can explain scientific concepts and write complex works, including dissertations, screenplays and functional computer codes. Claude de Loupy, head of the automatic text generation company Syllabs, said that ChatGPT offers a “really impressive” performance.
Compared with its predecessor, GPT-3, ChatGPT produces fewer absurd responses because it can ask for details. Its answers only become off the mark when users ask highly specific questions.
AI researcher Sean McGregor explained that chatbot technology had improved drastically over the years. According to McGregor, newer chatbots can act according to the history of inquiries and responses. In the past, they immediately forgot previous interactions.
Despite the improvements, de Loupy said ChatGPT still suffered from a weakness common to neural-mimicking programs — the inability to process meaning. Chatbots are unable to come up with reasonings for their choices.
McGregor said that OpenAI had designed ChatGPT so it would not be able to express opinions on various subjects. He had attempted to make the chatbot write a poem about ethics, and it refused to do the task.
OpenAI CEO Sam Altman addressed the re-sparked debate about the potential harms of AI via his Twitter account. He brought up the discussion of whether AI should act according to the intentions of its creators or the expectation of its users.
“The question of whose values we align these systems to will be one of the most important debates society ever has,” Altman added.
