Pineapple Express: Someone Is Wrong on the Internet?

[kermit4karate]

Exactly what actors and voice actors were striking over in recent years, while a lot of the public rolled their eyes at the notion that this was a real issue.

There are so many real issues related to AI, and most of them are unavoidable. Yet again, humanity is wading into a maelstrom of its own design. If not in the U.S. first, then someplace else. If not now, then next year. Sooner or later...miracles and nightmares.

 

[kermit4karate]

Most people don’t care until it directly affects them. The people cheering loudest about “AI” advancements now will be out of work and out of luck like the rest of us.

I use AI daily, multiple times, and I marvel at it. I'm also terrified of where it's headed, but how I feel about it and what it'll become, whether I use it now or not, has zero impact on its future. Whether you use it has zero impact too. That ship sailed years ago.

 

[Scribe]

The "lol, learn to code" crowd has gotten very quiet, now that low level knowledge of coding turns out to be one of the first things a large language model is able to replicate.

Uh for the record, I've been saying its an impending disaster for some time.

Thankfully right now AI is in its 'workslop' phase for most activities.

Give me 4 more years baby!

 

[Cadence]

Had someone turn in AI slop today. I let the class use it for help on parts they are stuck on but not the whole thing, say they should put things in their own words, and they need to say they used it if they did. In any case, the sentence along the lines of "Your output will include <brief description of what someone doing it would find, but without the numbers because it couldn't make the output in the language were using>" was kind of a give-away even if the perfect verbosity on the rest wasn't (I have trouble picturing an actual person not using at least a few ad-hoc abbreviations after writing some of those names enough).

 

[Scribe]

Had someone turn in AI slop today. I let them use it for help on parts they are stuck on but not the whole thing, should put things in their own words, and need to say they used it if they did. In any case, the sentence along the lines of "Your output will include <brief description of what someone doing it would find, but without the numbers because it couldn't make the output in the language we're using>" was kind of a give-away even if the perfect verbosity on the rest wasn't (I have trouble picturing an actual person not using at least a few ad-hoc abbreviations after writing some of those names enough).

Its going to be really unfortunate when an entire generation has no idea how to actually do something, and just gets by describing kind of sort of what they think they understand and hoping for the best.

 

[Cadence]

Its going to be really unfortunate when an entire generation has no idea how to actually do something, and just gets by describing kind of sort of what they think they understand and hoping for the best.

I just want them to try themselves before checking their answers (whether with a classmate or with AI) so that the in class exam doesn't chew them and their need-a-B-average scholarship up and spit them both out in tatters.

I think I might also start taking attendance next time in my upper-division classes. I used to only do it in the intro non-major mostly freshman course, and counted on interest or experience to motivate a lot more of them to show up. (I might not count it as part of the grade, but at least I'll know for sure who to hit the emergency support button for).

 

[The Firebird]

I use AI daily, multiple times, and I marvel at it. I'm also terrified of where it's headed, but how I feel about it and what it'll become, whether I use it now or not, has zero impact on its future. Whether you use it has zero impact too. That ship sailed years ago.

Teaching myself new things now, with LLMs where they are, is just so much faster than how it was before. Not as a replacement, but as an aide to a text or lecture series...to be able to pause right where you're confused, ask for clarification, context, whatever...work through some examples or ask for references to avoid hallucinations...it's night and day.

At the same time, it's eviscerating students who don't have much interest in learning or just want something to turn in for a grade.

 

[Scribe]

I just want them to try themselves before checking their answers (whether with a classmate or with AI) so that the in class exam doesn't chew them and their need-a-B-average scholarship up and spit them both out in tatters.

I failed out of my first year University, in spectacular "never been on his own has no clue how to live and got by in HS by just being smart enough" fashion. My Math final was a glorious fireball.

I dont envy educators right now, I doubt there is any kind of guidance that is going to actually land 100% correctly. My company has no clue how to deal with things, most dont if the 'workslop' articles are anything to go by.

I mean at least (some?) schools are limiting phones in class now? Yay for an absolutely obvious move?

Its all moving a bit faster than our highly evolved ... ahem brains can manage I think.

 

[kermit4karate]

Teaching myself new things now, with LLMs where they are, is just so much faster than how it was before. Not as a replacement, but as an aide to a text or lecture series...to be able to pause right where you're confused, ask for clarification, context, whatever...work through some examples or ask for references to avoid hallucinations...it's night and day.

At the same time, it's eviscerating students who don't have much interest in learning or just want something to turn in for a grade.

Me too, absolutely. I'm able to perform technical tasks for work in minutes that would have taken a team weeks to do a couple years ago. The increase in productivity is mind boggling.

Also terrifying and deeply disturbing? Yup, that too!

 

[Cadence]

Me too, absolutely. I'm able to perform technical tasks for work in minutes that would have taken a team weeks to do a couple years ago. The increase in productivity is mind boggling.

Also terrifying and deeply disturbing? Yup, that too!

I completely get it when the people using it know how to check the answers it gives (because they know the underlying ideas) and actually do check. I mean, I use it to double check all kinds of stuff I think I have worked out right. And can certainly imagine cases where I would ask it for things and then check them. (I've taken to running the assignments I give through it just to see what it gets right and wrong).

But I have no mercy for the author of a paper sent to a journal I edit for if it has hallucinated references. Or for a PhD student who asks AI for a short summary of a paper I told them to read and figure out and ask me about if they get stuck, and then thibks reading the summary is enough.

-----

It feels kind of like watching a pitcher go to the batting cages to get better. (As opposed to a batter going to them).

----

Also not eager to see what the influx of data centers due to our state's power costs and water supplies.