This is the correct take. To contrast the Terance Tao piece from earlier (https://news.ycombinator.com/item?id=46017972), AI research tools are increasingly useful if you're a competent researcher that can judge the output and detect BS. You can't, however, become a Terence Tao by asking AI to solve your homework.
So, in learning environments we might not have an option but to open the floodgates to AI use, but abandon most testing techniques that are not, more or less, pen and paper, in-person. Use AI as much as you want, but know that as a student you'll be answering tests armed only with your brain.
I do pity English teachers that have relied on essays to grade proficiency for hundreds of years. STEM fields has an easier way through this.
Andrej and Garry Trudeau are in agreement that "blue book exams" (I.e. the teacher gives you a blank exam booklet, traditionally blue) to fill out in person for the test, after confiscating devices, is the only way to assess students anymore.
My 7 year old hasn't figured out how to use any LLMs yet, but I'm sure the day will come very soon. I hope his school district is prepared. They recently instituted a district-wide "no phones" policy, which is a good first step.
Blue book was the norm for exams in my social science and humanities classes way after every assignment was typed on a computer (and probably a laptop, by that time) with Internet access.
I guess high schools and junior highs will have to adopt something similar, too. Better condition those wrists and fingers, kids :-)
I'm oldish, but when I was in college in the late 90s we typed a huge volume of homework (I was a history & religious studies double major as an undergrad), but the vast majority of our exams were blue books. There were exceptions where the primary deliverable for the semester was a lengthy research paper, but lots and lots of blue books.
Oh how I hated those as a student. Handwriting has always been a slow and uncomfortable process for me. Yes, I tried different techniques of printing and cursive as well as better pens. Nothing helped. Typing on a keyboard is just so much faster and more fluent.
It's a shame that some students will again be limited by how fast they can get their thoughts down on a piece of paper. This is such an artificial limitation and totally irrelevant to real world work now.
Maybe this is a niche for those low distraction writing tools that pop up from time to time. Or a school managed Chromebook that’s locked to the exam page.
That was how I took most of my school and university exams. I hated it then and I'd hate it now. For humanities, at least, it felt like a test of who could write the fastest (one which I fared well at, too, so it's not case of sour grapes).
I'd be much more in favour of oral examinations. Yes, they're more resource-intensive than grading written booklets, but it's not infeasible. Separately, I also hope it might go some way to lessening the attitude of "teaching to the test".
We had orals for graded programming assignments in graduate school. You had to present your solution to a panel and defend it. Some of my classmates really struggled with anxiety in front of the panel and took any question, however mild, as intense personal criticism.
Maybe this is a case for "learning styles", but it's probably logistically prohibitive to offer both options.
> My 7 year old hasn't figured out how to use any LLMs yet, but I'm sure the day will come very soon. I hope his school district is prepared. They recently instituted a district-wide "no phones" policy, which is a good first step.
This sounds as if you expect that it will become possible to access an LLM in class without a phone or other similar device. (Of course, using a laptop would be easily noticed.)
The phone ban certainly helps make such usage noticeable in class, but I'm not sure the academic structure is prepared to go to in-person assessments only. The whole thread is about homework / out of class work being useless now.
1. Corporate interests want to sell product
2. Administrators want a product they can use
3. Compliance people want a checkbox they can check
4. Teachers want to be ablet to continue what they have been doing thus far within the existing ecosystem
5. Parents either don't know, don't care, or do, but are unable to provide a viable alternative or, can and do provide it
We have had this conversation ( although without AI component ) before. None of it is really secret. The question is really what is the actual goal. Right now, in US, education is mostly in name only -- unless you are involved ( which already means you are taking steps to correct it ) or are in the right zip code ( which is not a guarantee, but it makes your kids odds better ).
> AI research tools are increasingly useful if you're a competent researcher that can judge the output and detect BS.
This assumes we even need more Terence Taos by the time these kids are old enough. AI has gone from being completely useless to solving challening math problems in less than 5 years. That trajectory doesn't give me much hope that education will matter at all in a few years.
The answer is you put the top mathematician in the world to do it, easy peasy.
“The argument used some p-adic algebraic number theory which was overkill for this problem. I then spent about half an hour converting the proof by hand into a more elementary proof, which I presented on the site.”
What’s the exchange rate for 30 minutes of Tao’s brain time in regular researcher’s time? 90 days? A year?
For that sort of task: no, Tao isn't all that much better than a "regular researcher" at relatively easy work. But the tougher the problems you set them at, the more advantage Tao will have.
... But mathematics gets very specialized, and if it's a problem in a field the other guy is familiar with and Tao isn't, they'll outperform Tao unless it's a tough enough problem that Tao takes the time to learn a new field for it, in which case maybe he'll win after all through sheer brainpower.
Yes, Tao is very very smart, but it's not like he's 100x better at everything than every other mathematician.
The 4000-5000 series Nikon Coolscans sell for about the same price they did 20 years ago because they still produce excellent scans and there’s nothing quite as good for that $1000-$15000 price out there.
The slide feeder is good but it's worth being aware that if you have slides mounted on cardboard (I had a lot of old family photos like this I used it for) it will often grab a couple at once. You can fix that by clipping eg a driver's licence in the right place to narrow the gap it pulls the slides through, but it will still need some manual supervision.
If you get one, have a look at VueScan on the software side - the original software needs (I think) a Windows XP virtual machine to drive it.
- DMSP satellites are up and measuring data
- These data will continue to be measured after Monday
- the government is discontinuing processing and public access to the data
- This will impact our capacity to predict hurricanes and monitor sea ice.
So, in learning environments we might not have an option but to open the floodgates to AI use, but abandon most testing techniques that are not, more or less, pen and paper, in-person. Use AI as much as you want, but know that as a student you'll be answering tests armed only with your brain.
I do pity English teachers that have relied on essays to grade proficiency for hundreds of years. STEM fields has an easier way through this.