Artificial intelligence is 'not human' and 'not intelligent' says expert, amid rise of 'AI psychosis'
Artificial Intelligence (AI) is not human and is not intelligent, an expert has told LBC as she urged people to be wary of snake oil salesmen.
Listen to this article
Professor Dame Wendy Hall, director of the Web Science Institute at the University of Southampton, told Shelagh Fogarty that while AI can be used for 'wonderful things' in the future, the technology has not yet reached a point where we can.
She said: "I think we all need to calm down a bit.
"There are a lot of snake oil salesmen out there, there are a lot of people out there who try and sell you things, but we have to understand this stuff is not human, it is not intelligent.
"You could do all sorts of wonderful things with it and over time, I think we will learn how to use this sort of tool for therapy, but we aren't there yet."
Her comments come after Microsoft boss Mustafa Suleyman warned of the number of people who are suffering from 'AI psychosis'.
AI psychosis is a non-clinical term that refers to somebody losing touch with reality after heavy use of tools such as Chat GPT.
Some believe that the AI they are talking to has developed consciousness and has human emotions.
Read more: Police forces to use artificial intelligence to investigate grooming gangs
Many web users who have spent the last five months getting to know GPT-4o say its replacement - GPT-5 – seems less friendly and responsive.
There are around 17,000 members of the MyBoyfriendIsAI Reddit group, where people share their experiences of their close “relationships” with AI.
The head of AI at Microsoft said that there is 'zero evidence' of AI consciousness and that developing an unhealthy attachment to a chatbot can affect anyone.
He wrote on X: "There's zero evidence of AI consciousness today. But, if people just perceive it as conscious, they will believe that perception as reality.
"Even if the consciousness itself is not real, the social impacts certainly are."
'Rapidly destroying the internet'
One of the potential issues with AI is that it is being trained on information found online, but, Dame Hall said that "we are rapidly destroying the internet" with the kind of content being put out.
She has urged people to be more cautious about trusting what is being said by AI and to ask themselves whether they can trust the information they're reading.
She said: "I'm much more worried about that because we're putting a lot of rubbish out there.
"People read stuff from Chat GPT, put it out without knowing whether it's right or wrong or indifferent.
"And then that's used to train the new versions, the new models. So we are making the Internet garbage in, garbage out sort of stuff."