The thing is that LLMs are actually really great at helping you learn - they can help to form connections between ideas and surface new knowledge much, much faster than traditional research or the now worthless search engine.
You just have to keep in mind that you need to externally validate everything. If you can keep this in mind then LLMs really are a great way of lighting the way towards some obscure topic you’d like to know more about, and can provide very useful guidance on solving problems.
They can lead you to water but you need to take the drink yourself.
I hear you. I’d still be hesitant to let school age kids learn with an LLM companion. If the grownups think they’re talking with a sentient gigabyte, I think the danger is too great to expose kids to this. Which brings me to my big picture opinion: the general public doesn’t need to have access to most of these models. We don’t need to cook polar bears alive to make 5 second video memes, slop, or disinformation. You can just read your emails. No one needs ChatGPT plan their next trip. No one should consider an LLM a substitute for a trained therapist. There are good applications in the field of accessibility, probably medical as well. The rest can stay in a digital lab until they’ve worked out how not to tell teenagers to kill themselves, not to eat rocks to help your digestion, or insert any other bullshit so-called AI headline you have read recently here. It’s not good for people, the environment, and it’s forming a dangerous bubble that will have shades of subprime mortgages 2007/8 when it bursts. The negatives outweigh the positives.
You need to be able to think critically to use an LLM as an effective research tool and children in schools fuckjng suck at that, so I agree with you.
But now that I think about it - they could be a good way to actually teach critical thinking.
We NEED to be raising a generation of relentless critical thinkers if we don’t want to slip into facism, and we are doing an absolutely shit job of it.
Agreed.
The thing is that LLMs are actually really great at helping you learn - they can help to form connections between ideas and surface new knowledge much, much faster than traditional research or the now worthless search engine.
You just have to keep in mind that you need to externally validate everything. If you can keep this in mind then LLMs really are a great way of lighting the way towards some obscure topic you’d like to know more about, and can provide very useful guidance on solving problems.
They can lead you to water but you need to take the drink yourself.
I hear you. I’d still be hesitant to let school age kids learn with an LLM companion. If the grownups think they’re talking with a sentient gigabyte, I think the danger is too great to expose kids to this. Which brings me to my big picture opinion: the general public doesn’t need to have access to most of these models. We don’t need to cook polar bears alive to make 5 second video memes, slop, or disinformation. You can just read your emails. No one needs ChatGPT plan their next trip. No one should consider an LLM a substitute for a trained therapist. There are good applications in the field of accessibility, probably medical as well. The rest can stay in a digital lab until they’ve worked out how not to tell teenagers to kill themselves, not to eat rocks to help your digestion, or insert any other bullshit so-called AI headline you have read recently here. It’s not good for people, the environment, and it’s forming a dangerous bubble that will have shades of subprime mortgages 2007/8 when it bursts. The negatives outweigh the positives.
You need to be able to think critically to use an LLM as an effective research tool and children in schools fuckjng suck at that, so I agree with you.
But now that I think about it - they could be a good way to actually teach critical thinking.
We NEED to be raising a generation of relentless critical thinkers if we don’t want to slip into facism, and we are doing an absolutely shit job of it.