Artificial Intelligence is being banked on by scientists to help them in the battle against cancer.
However an Australian breakfast radio duo have tested the ChatGPT AI program with a very basic question – only for it to spectacularly bungle the answer.
On Thursday morning’s The Kyle & Jackie O Show, hosts Kyle Sandilands and Jackie O Henderson decided to put the ChatGPT through its paces live on-air – and the results left them shocked.
The program, which was created by the company Open AI, was not able to correctly answer the simple question: ‘How many Rs are in the word “strawberry”?’
When asked the question, the app claimed there are two Rs in the word – rather than the correct answer of three.
The program is meant to learn from its interactions with and corrections by users, however the chatbot still made the same error even after being corrected multiple times.
Kyle, 53, quipped, ‘We’re not going to be curing cancer any time soon with this thing are we?’
After playing a viral clip of the app making the spelling mistake, the hosts asked their show director Bruno Bouchet to test it while they were broadcasting.
‘How many Rs are in strawberry?’ asked Bruno, before the chatbot replied, ‘there are two Rs.’
An Australian breakfast radio duo have tested ChatGPT with a simple question which it repeatedly answers incorrectly. Pictured: radio host Kyle Sandilands
The studio then gasped in shock as others began laughing.
Bruno then attempted to test the AI’s learning capabilities, saying, ‘It sounds to me like there are three.’
‘I see we’re you’re coming from,’ said the robotic voice, before making a wildly inaccurate claim.
‘It might feel like there are more R sounds, but in the word “strawberry” there are only two Rs – one after the T and one at the end of the word.’
After their attempts to educate the AI failed, the radio stars then complained about the hype around such programs.
‘You stupid piece of s**t,’ Kyle said, before adding ‘I thought this thing would cure cancer.’
The test comes just months after Scottish scientists revealed they have developed an Artificial Intelligence system that can interpret cancer samples and predict how a tumour might progress.
‘How many Rs are in strawberry?’ asked a show producer, before the chatbot replied, ‘there are two Rs.’ Pictured: host Jackie O Henderson
The team hopes the discovery could give doctors an AI ‘second opinion’ and lead to faster, more accurate diagnosis.
An international team of AI specialists and cancer scientists, led by researchers from the University of Glasgow, have developed a new system, which they call histomorphological phenotype learning (HPL).
They took thousands of high-resolution images of tissue samples of lung cancer patients and collected data on how the cancers progressed.
The test comes just months after Scottish scientists revealed they have developed an Artificial Intelligence system that can interpret cancer samples and predict how a tumour might progress. Pictured: stock
Next, they developed an algorithm to analyse the images and spot patterns based solely on the visual data in each slide.
The algorithm broke down the slide images into thousands of tiny pieces, each representing a small amount of human tissue.
When the team added slides from more cancer patients to the system, it was capable of correctly distinguishing between their features.
Once the algorithm had identified patterns in the samples, the researchers used it to analyse links between the samples and the patients’ outcomes stored in the database, including how long patients lived after having cancer surgery.
Another recent study claimed AI was better at detecting prostate cancer than doctors.
By Daily Mail Online, August 29, 2024
