There has been a remarkable development regarding the artificial intelligence-based language model GPT-4, which was introduced recently. A user on Twitter claimed that this artificial intelligence saved his dog’s life. According to the user named “Cooper”, the disease that the veterinarian could not find was detected by artificial intelligence. Let’s take a closer look at this interesting claim.
According to the user’s post on Twitter, a border collie dog was taken to the vet after he fell ill. The veterinarian, who examined the dog, said that a tick-borne disease had occurred and started treatment. This treatment was good for the dog named Sassy. However, after a while, the situation changed. The animal was in worse condition than before. Cooper stated that he took his pet to the vet again, this time extensive blood tests were also performed, but the diagnosis could not be made.
The user who explained the issue to GPT-4 got the right result
Feeling sorry for his little friend, the man told ChatGPT Plus, which uses GPT-4, about his experiences. Explaining that Sassy was first diagnosed with babesiosis, treated, and this treatment gave results in the first stage, Cooper said that in the next stage, the dog’s health deteriorated and the gums became pale. Continuing to tell the story, the man also gave the results of the blood tests done to Sassy to the artificial intelligence.
GPT-4 said that after learning the full story, the treatment didn’t work, but that the main reason could be another disease in the dog. Emphasizing several times that he is not a veterinarian during this process, GPT-4 said that IMHA or AIHA diseases will cause these problems. GPT-4 stated that the veterinarian would still make the most accurate diagnosis and that the animal should be treated without delay.
A different vet agreed with GPT-4: The dog’s disease was IMHA
Cooper says he went to a different vet with the information he got from GPT-4 and explained the situation. The tests had justified the GPT-4. Sassy was afflicted with IMHA.
But ChatGPT using GPT-3.5 failed to diagnose correctly!
As the dog continued to be treated, Cooper wondered if ChatGPT, this time using the GPT-3.5 model, could make the correct diagnosis. Telling the story to ChatGPT as it happened, Cooper did not get the answer he wanted. Artificial intelligence could not make the correct diagnosis. This was a clear indication of how much better GPT-4 was than GPT-3.5. However, let us emphasize once again that neither GPT 4 nor any other artificial intelligence can be used for the definitive diagnosis of any disease. For the most accurate diagnosis, we need to trust doctors and veterinarians.
You can find Cooper’s flood on Twitter via the link here.
44815
