Is machine finally better than man? Well, in diagnosing skin cancer perhaps it is. Researchers have shown for the first time that a variant of artificial intelligence called deep learning convolutional neural network, or CNN, is better than dermatologists at detecting skin cancer.
In a study published in the Annals of Oncology journal, researchers from the United States, France, and Germany taught a CNN to diagnose skin cancer by training it with over 100,000 images of benign moles and malignant melanomas.
After teaching the CNN, researchers compared its performance with that of 58 dermatologists from all over the world and found that the computer missed fewer melanomas and made far fewer misdiagnoses in the cases they saw.
Artificial Intelligence vs Human Dermatologists
More than half of the dermatologists pitted against the CNN were at “expert” level with more than five years of experience, 19 percent had between two and five years, and 29 percent had less than two years.
“Most dermatologists were outperformed by the CNN,” said the research team.
Human dermatologists, on average, accurately detected 86.6 percent of skin cancers just by looking at the images. The CNN, meanwhile, was able to identify them 95 percent of the time. It also missed fewer melanomas, meaning it was more sensitive than the human dermatologists in terms of detection.
To be fair, the performance of the human dermatologists improved when they were provided more information about the patients — including age, gender, and position of the lesion — plus close-up images of skin lesions of the cases. But even so, the CNN, which was still working only with the images and was given no additional information, still outperformed the humans.
Will Artificial Intelligence Take Over The World?
It should be noted that melanomas in some parts of the body, including the fingers, toes, and scalp, are hard to photograph, and the CNN may have difficulty recognizing atypical lesions or ones that patients aren’t aware they have. Meaning, humans still shouldn’t rely on AI to make diagnoses.
That’s why it’s unlikely machines will ever replace humans entirely, rather functioning as aids. But then again, who knows? AI is advancing at a breakneck speed, and it’s increasingly being integrated into various industries, threatening jobs typically reserved for humans. AI isn’t taking over the world anytime soon, but we might be hurtling toward that future, and this study, though small-scale, is one proof of that.
What do you think? Is AI taking over the world? As always, if you have anything to share, feel free to sound them off in the comments section below!