Hacker News Reader

Comment by xnx | Hacker News Reader

> GenAI is a threat to student learning and wellbeing.

This blanket dismissal is not going to age well, and reads like a profession lashing out.

With the right system prompt, AI can be a patient, understanding, encouraging, non-judgemental tutor that adapts and moves at the student's pace. Most students can not afford that type of human tutor, but an AI one could be free or very affordable.

"How AI Could Save (Not Destroy) Education" (https://www.youtube.com/watch?v=hJP5GqnTrNo) from Sal Khan of Khan Academy

View on HN

> AI can be a patient, understanding, encouraging, non-judgemental tutor

Groan... no it can't. It can simulate all those things, but at the moment, "AI" can't be patient, understanding, and whether judgemental or non-judgemental.

OK it can be encouraging. "You're one good student, $STUDENT_NAME!" (1).

1) https://www.youtube.com/watch?v=jRPPdm09xZ8

I can say the exact same thing about you or anybody else. You can’t be patient, understanding, encouraging, non-judgmental tutor. You can only simulate it.

I really can’t understand why people don’t understand this. What am I missing?

Geezus freaking christ.

Now is that a simulation of someone who thinks he's responding to a cretin... or actually the feelings of someone who thinks he's talking to a cretin?

Philosophical zombies are supposed to be a thought experiment to demonstrate that solipsism and nihilism are stupid, not a rhetorical device to equate human minds to linear algebra statistical parrots.

> What am I missing?

You are missing what this "AI" is: a guessing system based on some patterns.

Every argument against AI always falls back to some weird reductionism that simply does not match up with reality as if it's a gotcha. I pity the fool that has such a poor working model of reality based on whatever internal pessimistic biases you may have.

> You are missing what this "AI" is: a guessing system based on some patterns.

You just described the majority of your daily thoughts.

Whether the AI is patient, understanding, etc., is entirely up to the person interacting with it to decide. Just like they decide this when interacting with people. You can never know the internal state of the other in a conversation so it is up to you to model it and if modeling it is best done with human metaphors then use human metaphors.

Most student can not afford the expertise necessary to have AI patient etc.

I think the original phrase was made with the assumption "as it is right now".

I do share concerns of undersigned, even though don't necessarily agree with all statements in the letter.

If you are using the most commonly available AI and have an average ability of perfecting a search term, right now AI is not a particularly useful tool in learning anything. It is far too inaccurate to learn anything challenging. The key term here is could, and yes it is possible but there is nothing yet to say we shall get there.

My experience in higher education is that students use AI for one of two things:

1. To do the homework because they view classes and grades as a barrier to their future instead of preparation for such.

2. In place of a well crafted query in an academic database.

It's not a blanket dismissal, it's a fact in context. It should read like a profession lashing out - that's what it is.

AI has enormous upsides and enormous downsides. The "you're going to look so dumb in the future" dismissal is lazy. Inevitability does not make something purely beneficial.

It's a fallacious line of thinking that's disappointingly common in tech-minded people (frequently seen in partnership with implications that Luddites were bad or stupid, quotes from historical criticisms of computers/calculators, and other immature usage of metaphor).

I'd respect the statement more if it acknowledged that AI had some benefit, or potential benefit in the future, but they did not want to use it currently.

Maybe if we move from LLMS to real AI it will have benefits.

Reply on news.ycombinator.com