thing we have no way of knowing how it works, therefore no way of relying on it
thing that helps you do something you then have to do anyway by yourself (if you want to learn something from generative model output you still need to fact check it)
vague promise it will lead to anything useful in the future
You don’t always have to know how it works to rely on it. Most people could not tell you how a computer works but they are able to do better work.
We can verify that it’s better in some tasks than people. E.g give doctors and the AI 1000 MRI scans of potential cancer patients and it can determine it more accurate than doctors. So there they already are a help.
It’s already used in advancing different fields, for example reading texts of ancient burned scrolls without opening the scroll since that would break them.
And with learning, yeah like books, they help you to learn faster but are not a requirement. Same here I can learn much faster now. But I will verify what it tells me.
—
But I’m not sure if all of that outweighs the shit. 💩
The genie is already out of the bottle, no putting it back.
You don’t always have to know how it works to rely on it. Most people could not tell you how a computer works but they are able to do better work.
We can verify that it’s better in some tasks than people. E.g give doctors and the AI 1000 MRI scans of potential cancer patients and it can determine it more accurate than doctors. So there they already are a help.
I’d really prefer if my doctor knew why they say I have cancer!
That would be nice but as the “proud owner of medical issues” it’s much more often: “You have this, we don’t know why you have it, this is how we can manage it”.
You still want your doctor to be knowledgeable of course, but you also want them to use the best tools at their disposal. Most of them probably couldn’t tell you how an mri machine works exactly either.
So the pros are
You don’t always have to know how it works to rely on it. Most people could not tell you how a computer works but they are able to do better work.
We can verify that it’s better in some tasks than people. E.g give doctors and the AI 1000 MRI scans of potential cancer patients and it can determine it more accurate than doctors. So there they already are a help.
It’s already used in advancing different fields, for example reading texts of ancient burned scrolls without opening the scroll since that would break them.
https://www.theguardian.com/science/2025/feb/05/ai-helps-researchers-read-ancient-scroll-burned-to-a-crisp-in-vesuvius-eruption
But also medicine creation etc.
And with learning, yeah like books, they help you to learn faster but are not a requirement. Same here I can learn much faster now. But I will verify what it tells me.
— But I’m not sure if all of that outweighs the shit. 💩 The genie is already out of the bottle, no putting it back.
I’d really prefer if my doctor knew why they say I have cancer!
That would be nice but as the “proud owner of medical issues” it’s much more often: “You have this, we don’t know why you have it, this is how we can manage it”.
You still want your doctor to be knowledgeable of course, but you also want them to use the best tools at their disposal. Most of them probably couldn’t tell you how an mri machine works exactly either.