Just like Biden saying who counts the votes is the most important variable, it's who is supplying the training data for the AI that ultimately matters.
Also, a lot of these systems are trained on old data. I wouldn't be surprised if 2022 is too new for them to know about it. AI doesn't know everything that happens in real time, nor is it truly capable of learning. One the training has finished, it's only capable of very basic adjustments to the way it processes the data.
So western AI medical programs are going to cause more malpractice than Doug from Scubs.
While we'll have a ton of articles from the mainstream going 'why are Asians living so long' because they don't care if their AI is racist or non pc, that's a valuable function.
Even this chart is BS for a bunch of reasons (like trusting covid data from NY/CA).
The biggest problem is they did nothing to match the groups on health. Other studies have shown non-covid-vaccinated group as being overall significantly more likely to die, so much of this difference in the graph may just be due to the groups being different health to begin with.
There may be any number of reasons for that, for example people in hospice or with a cancer death sentence may avoid the vaccine since they've made peace with dying and this study makes no attempt to correct for that. Ultimately these are self-selected groups and are not directly comparable like this paper did.
The authors intended to show a big benefit from the vaccine:
Second, persons with undiagnosed infection are misclassified as having no previous COVID-19 diagnosis; however, this misclassification likely results in a conservative bias
Non-biased authors would identify "undiagnosed infection are misclassified as having no previous COVID-19 diagnosis" as a limitation, but adding "however that's ok because it still showed what we wanted" shows their intentions; they're ok with the difference being bigger, but not smaller, because this is what they intended to show.
I would not ascribe such agency to the chatbots. If the mainstream media is spewing medical misinformation, then the chatbots will parrot that misinformation.
I don't understand the drive to anthropomorphise these text completion algorithms. They are not 'lying' because they don't have any free will of their. They can't 'lie'. They are simply producing patterns of text based on what they have been trained to say. Stop making these things seem more important that they actually are by treating as if they were conscious beings.
It's not lying per se. It's just repeating what it has been taught.
Just like Biden saying who counts the votes is the most important variable, it's who is supplying the training data for the AI that ultimately matters.
AI doesn't know anything at all, it's just a giant auto-complete function.
Also, a lot of these systems are trained on old data. I wouldn't be surprised if 2022 is too new for them to know about it. AI doesn't know everything that happens in real time, nor is it truly capable of learning. One the training has finished, it's only capable of very basic adjustments to the way it processes the data.
So western AI medical programs are going to cause more malpractice than Doug from Scubs.
While we'll have a ton of articles from the mainstream going 'why are Asians living so long' because they don't care if their AI is racist or non pc, that's a valuable function.
Even this chart is BS for a bunch of reasons (like trusting covid data from NY/CA).
The biggest problem is they did nothing to match the groups on health. Other studies have shown non-covid-vaccinated group as being overall significantly more likely to die, so much of this difference in the graph may just be due to the groups being different health to begin with.
There may be any number of reasons for that, for example people in hospice or with a cancer death sentence may avoid the vaccine since they've made peace with dying and this study makes no attempt to correct for that. Ultimately these are self-selected groups and are not directly comparable like this paper did.
The authors intended to show a big benefit from the vaccine:
Non-biased authors would identify "undiagnosed infection are misclassified as having no previous COVID-19 diagnosis" as a limitation, but adding "however that's ok because it still showed what we wanted" shows their intentions; they're ok with the difference being bigger, but not smaller, because this is what they intended to show.
I would not ascribe such agency to the chatbots. If the mainstream media is spewing medical misinformation, then the chatbots will parrot that misinformation.
I don't understand the drive to anthropomorphise these text completion algorithms. They are not 'lying' because they don't have any free will of their. They can't 'lie'. They are simply producing patterns of text based on what they have been trained to say. Stop making these things seem more important that they actually are by treating as if they were conscious beings.
Like everything else in Big Tech, MLMs have been brainwashed into Progressive, aka Rainbow-Marxist, ideology.