Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Failing grades soar with AI usage, dwindling math skills in Berkeley CS classes

dailycal.org

Post reply on HN
Print view View original

761–770 of 828 posts

Re: Failing grades soar with AI usage, dwindling math skills in Berkeley CS classes

#761

Post by numeri »

calepayson wrote: post #708
rahimnathwani wrote: post #643

Earlier quoted context omitted.

So all groups outsourced their thinking to an LLM. Will they have learned anything they can apply to different types of projects in future?

In your opinion where is the line on where LLMs are useful/harmful to learning? In my school projects I've found them super useful for working with libraries. In the past it felt like the theory I learned was pretty low impact. 80% of my time was spent learning the quirks of a library. Now it feels like I can take theory and iterate over a ton of different solutions without having to worry about learning whatever lib…

I would just warn that you may not be able to recognize what is worth learning at your stage.

Intuition for library design and the architecture of software packages/external APIs is something you can only learn by doing.

Re: Failing grades soar with AI usage, dwindling math skills in Berkeley CS classes

#763

Post by WorldMaker »

jawilson2 wrote: post #620
WorldMaker wrote: post #613

Earlier quoted context omitted.

The current trend seems to be switching the priority order of calls and texts among many of us. I feel like a call should be scheduled, preferably 3+ days out, and preferably with an agenda attached. (Same rules I feel about any sort of meeting.) But a direct text (non-group chat, just to me) is a priority. Group chats get that 1-2 days middle ground.

I know that's the trend, but it is backwards to me. Like UDP vs TCP. If you need an immediate answer for something, why send a one-way communication where you have no idea whether the person on the other end A) received it, and B) acted on it. A 15 second phone call accomplishes this, whereas if I text you it could be hours, unless you immediately respond.

Most text message apps have real time read receipts for A and "actively writing" indicators that imply some of B. (The things that we invented for "instant messaging" decades ago are finally mostly back in vogue in text messaging.) Text messages have "reactions" like Thumbs Up that very quickly say "I will action this shortly." For me that's more TCP, the classic ACK receipt then action.

A 15 second phone call probably goes straight to my voicemail and there's no read receipts if I have read the voicemail transcription yet, much less tried to listen to it. It might be hours until I'm in a quiet enough space to try to listen to it, because I don't carry headphones with me most of the time, don't listen to speakerphone in public, and the "phone speaker" on an iPhone is generally hard to hear for me, no matter how I awkwardly position the phone. The transcriptions are usually not good enough to action directly so I do usually need to do that dance of find a way to listen to it. Calls to me are a UDP hole with fewer acknowledgements and a lot of inconvenience.

Unless you've prescheduled a call with me, in which case maybe I do answer and maybe it is quick enough to be a 15 second phone call. But the easiest way to schedule that call with me is going to be to text me "Hey can I call in an hour about X?" and I'll ACK it soon after, and probably then likely spend 45 minutes looking for a quiet place to take a private phone call.

But, yeah, different perspectives for different sorts of people. Phone calls are taxing to me and the real world is a loud place and I don't carry headphones and I like to control the environment in which I try to answer a phone call, but also I find finding such quiet environments stressful enough to want to schedule them ahead of time.

Re: Failing grades soar with AI usage, dwindling math skills in Berkeley CS classes

#765

Post by WorldMaker »

II2II wrote: post #707
WorldMaker wrote: post #613

Earlier quoted context omitted.

The current trend seems to be switching the priority order of calls and texts among many of us. I feel like a call should be scheduled, preferably 3+ days out, and preferably with an agenda attached. (Same rules I feel about any sort of meeting.) But a direct text (non-group chat, just to me) is a priority. Group chats get that 1-2 days middle ground.

Wouldn't it depend upon context? If it is an emergency, it is a voice call. It is both immediate and conveys urgency. If it is something that you need to talk through, it is a scheduled voice call. Asynchronous communications may demonstrate respect for a person's time since it does not (need to) interrupt them in the moment, but the inefficiency results in a disrespectful waste of time for bidirectional conversation…

If it's an emergency I want it in text first because I read faster than I listen to voicemails, and can do so in more spaces/contexts/environments (say, at a loud concert or eating in a restaurant), and I trust a person to write the message better than the voicemail system will transcribe a call. Certain emergency keywords sent in a text will especially alert me faster than a call would. (I have call notifications as practically turned off as possible in current operating systems [way too much phone call spam], but a measured set of notification levels for text notifications.)

I think even for emergency situations the relationship between voice calls and text messages is flipping. Text messages are immediate and can convey urgency. Phone calls are for private, quiet spaces, which take time to find (or schedule). With the death of the private phone booths in public spaces, phone calls are inconvenient to take almost everywhere now outside of one's home, but urgent text messages can be read and even acted upon just about anywhere.

Re: Failing grades soar with AI usage, dwindling math skills in Berkeley CS classes

#766

Post by sourcecodeplz »

amiga386 wrote: post #669
HanayamaTriplet wrote: post #42

Earlier quoted context omitted.

If the removal of standardized testing in 2021 was the real reason, then why is there a sudden spike of failure rates happening right now?

https://senate.ucsd.edu/media/740347/sawg-report-on-admissio... Please see the graph "Growth of the Math 2 Population by Major (2019-2024)". UCSD's Math 2 class is remedial high-school level maths. It has grown from under 100 students in 2016-2020, to more and more people each year starting from 2021. UCSD tested the people who took this class, and 25% of them could not answer the question "Fill in the box: 7 + 2 = […

Another one with %21 fail: "Sarah had 9 pennies and 9 dimes. How many coins did she have in all?"

Re: Failing grades soar with AI usage, dwindling math skills in Berkeley CS classes

#767

Post by lukan »

brewdad wrote: post #756
lukan wrote: post #658

Earlier quoted context omitted.

Are you sure he is sending automated messages, or used LLM to polish up some writing? Would be a difference to me, but .. still weird.

No message between friends should ever need “polishing” unless you are trying to deliver really bad news in the softest way possible.

To be honest, especially then I would feel offended.

Re: Failing grades soar with AI usage, dwindling math skills in Berkeley CS classes

#769

Post by ygouzerh »

jventura wrote: post #241

CS Professor here: just yesterday I did the discussion of a course projects' (Parallel Computing), and one of the three groups that I did yesterday have clearly gone the ChatGPT way. They couldn't even understand the choices the LLM made regarding the architecture, etc. The way to "catch" these students is similar to what we did in the past when students copied from other students which is "to give them rope to hang"…

How do you see impact of using LLM on the teaching content? (genuine curiosity)

In the industry, almost everyone here is doing the exact same thing, outsourcing more and more of their thinking to LLM: so even if students learned how to manually write code, they will probably loose it later on (happening to us already, mostly can be seen when people are working on new solutions/frameworks... they now have the same issue than students).

Re: Failing grades soar with AI usage, dwindling math skills in Berkeley CS classes

#770

Post by fransje26 »

hirvi74 wrote: post #662
fransje26 wrote: post #219

Earlier quoted context omitted.

> but I do see I'm a lot "lazier", even stuff that used to be routine when I started coding now feel heavy. Not getting that quick dopamine hit the LLMs give you.. Some say you can re-train your system to get back the dopamine hits you used to get from other things, like the enjoyment of the "old fashioned" manual coding and math. Getting there is hard work. And YMMV.

I just do things manually and ask LLMs to check my work. That seems to be working great for me. I had the most Russian of Russian bosses when I was in college. My first day on the job he so eloquently stated, "I am not your mother. Do not come to me with problems. Come to me with solutions. I want to know what you tried and what did not work." His advice has served me well in many areas of life too. I try my best to…

What I would like to do is double model post-check, with a form of "debate", to better catch edge cases.

Unfortunately, I haven't found a way to set that up as I envisioned it, for the time being.

Post reply on HN