3 ms·

> Instead the bottom-up (neural networked) AI seems to be winning out. What do you mean by ... winning out? It seems to have made one novel diagnosis, which is

> Instead the bottom-up (neural networked) AI seems to be winning out. What do you mean by ... winning out? It seems to have made one novel diagnosis, which is great for the individual in question. But how many times has it diagnosed something completely absurd? What's the ratio of signal to noise here? If it's anything like trying to use it to solve difficult engineering problems it must have hallucinated turbocancer hundreds of times over.

famouswaffles 3y ago
>What do you mean by ... winning out? Symbolic AI has been less performant than NN AI for decades now and by a large margin. Language, speech, images etc. >It seems to have made one novel diagnosis Ok so let's get this out the way. GPT-4 is good at diagnosis https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10425828/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10425828/ >Six patients 65 years or older (2 women and 4 men) were included in the analysis. The accuracy of the primary diagnoses made by GPT-4, clinicians, and Isabel DDx Companion was 4 of 6 patients (66.7%), 2 of 6 patients (33.3%), and 0 patients, respectively. If including differential diagnoses, the accuracy was 5 of 6 (83.3%) for GPT-4, 3 of 6 (50.0%) for clinicians, and 2 of 6 (33.3%) for Isabel DDx Companion