Oxford University’s AI Professor Mike Wooldridge is suggesting everyone to be careful with chatbots like ChatGPT. He said that sharing private information or having personal talks with these chatbots is not a good idea because whatever you say helps make future versions smarter.
Wooldridge mentioned that chatbots won’t always give fair answers as they tend to tell you what you want to hear. He cautioned everyone with all these during his the Royal Institution Christmas lectures, where he answered a plethora of questions of listeners about how AI works.
He talked about how machines learn to translate languages and how chatbots do their job. He tried to answer intense questions such as can AI be just like humans.
Wooldridge said looking for feelings or understanding in AI is a waste of time. AI does not have feelings like empathy or sympathy. It is just programmed to say things that make you happy.
He further warned saying that if you type something into ChatGPT, you should assume it will be used in future versions. And once you tell something to ChatGPT, it’s tough to take it back because of how the AI works.
The people behind ChatGPT, called OpenAI, say they added a feature in April to turn off chat history. If you start a chat with history turned off, what you say won’t be used to teach the AI.
Wooldridge brought some robot friends with him to show what robots can and can’t do today. These lectures at the Royal Institution Christmas started way back in 1825 to teach young people about science and are the oldest science TV series. Over the years, the lectures have been delivered by renowned legends including Nobel Prize winner Sir David Attenborough.
