Present the results of your exercises (in person) in front of someone. Or really anything in person.
A big downer on the online/remote Initiatives for learning but actually an advantage for older Unis that already have existing physical facilities for students.
This does however also have some problems similar to code interviews .
> Present the results of your exercises (in person) in front of someone
I would not be surprised if we start to see a shift towards this. Interviews instead of written exams. It does not take long to figure out whether someone knows the material or not.
Personally, I do not understand how students expect to succeed without learning the material these days. If anything, the prevalence of AI today only makes cheating easier in the very short term -- over the next couple years I think cheating will be harder than it ever was. I tried to leverage AI to push myself through a fairly straightforward Udacity course (in generative AI, no less), and all it did was make me feel incredibly stupid. I had to stop using it and redo the parts where I had gotten some help, so that my brain would actually learn something.
But I'm Gen X, so maybe I'm too committed to old-school learning and younger people will somehow get super good at this stuff while also not having to do the hard parts.
The main challenge is that most (all?) types of submissions can be created with LLMs and multi-model solutions.
Written tasks are obvious, writing a paper, essay or answering questions is part of most LLMs advertised use-cases. The only other thing was recorded videos, effectively recorded presentations, thanks to video/audio/image generation that probably can be forged too.
So the simple solution to choose something that an "LLM can't do" is to choose something were an LLM can't be applied. So we move away from a digital solution to meatspace.
Assuming that the goal is to test your knowledge/understanding of a topic, it's the same with any other assistive technology. For example, if an examiner doesn't want you[1] to use a calculator to solve a certain equation, they could try to create an artificially hard problem or just exclude the calculator from the allowed tools. The first is vulnerable to more advanced technology (more compute etc.) the latter just takes the calculator out of the equation (pun intended).
[1]: Because it would relieve you of understanding how to evaluate the equation.
That was an interesting read. The core issue unfortunately appears to boil down to being a deeply complex of net of trust and risk distribution.
The issue being a "human"/societal problem, instead of technology, makes me wonder if this could be slowly changed over a few generations. The amount of momentum required would be quite high tough.
1. Make the student(s) randomly have to present their results on a weekly basis. If you get caught for cheating at this point, at least in my uni with a zero tolerance policy, you instantly fail the course.
2. Make take home stuff only a requirement to be able to participate in the final exam. This effectively means cheating on them will only hinder you and not affect your grading directly.
3. Make take home stuff optional and completely detached from grading. Put everything into the final exam.
My uni does a mix them on different courses. Especially two and three though have a significant negative impact on passing rates as they tend to push everything onto one single exam instead of spread out work over the semester.
> in that every software engineer now depends heavily on copilots
That is maybe a bubble around the internet. Ime most programmers in my environment rarely use and certainly aren't dependent on it. They do also not only do code monkey-esque web programming so maybe this is sampling bias though it should be enough to refute this point.
Came here to say that. It’s important to remember how biased hacker news is in that regard. I’m just out of ten years in the safety critical market, and I can assure you that our clients are still a long way from being able to use those. I myself work in low level/runtime/compilers, and the output from AIs is often too erratic to be useful
Documentation search I might agree, but that wasn’t really the context, I think. Code reviews is hit and miss, but maybe doesn’t hurt too much. They aren’t better at writing good tests than at writing good code in the first place.
are there any downsides to adding "bad tests" though? as long as you keep generated tests separate, it's basically free regression testing, and if something meaningfully breaks on a refactor, you can promote it to not-actually-slop
I would say that the average Hacker News user is negatively biased against LLMs and does not use coding agents to their benefit. At least what I can tell from the highly upvoted articles and comments.
Im on the core sql execution team at a database company and everyone on the team is using AI coding assistants. Certainly not doing any monkey-esque web programming.
> everyone on the team is using AI coding assistants.
Then the tool worked for you(r team). That's great to hear and maybe gives some hope for my projects.
It has just mostly been more of a time sink than an improvement ime though it appears to strongly vary by field/application.
> Certainly not doing any monkey-esque web programming
The point here was not to demean the user (or their usage) but rather to highlight how developers are not being dependent on LLMs as a tool. Your team presumably did the same type of work before without LLMs and won't become unable to do so if there were to become unavailable.
That likely was not properly expressed in the original comment by me, sorry.
I'm hosting it for our team at a public institute, we are strongly supportive of OSS and have interest in keeping our data on premise.
Team of <10 though so hosting is trivial with NixOS. We also have almost no money available for purchasing software so official self-hosted bitwarden was not an option unfortunately (if we had money, that would've been the way to go).
Yep, there are still server CPUs with only P-cores.
They are a bit expensive but I wouldn't expect them to drop these skews in the long term for HPC & compute bound workloads. My guess is that diamond rapids will also have some P-skews and maybe AP skews.
Iirc that is mostly an issue of the scheduling and not the construction itself. The operator is also known to have reliability problems with many train types (not necessarily the stops or tracks themselves).
> Managing multiple Linux servers usually means juggling terminal windows and copy-pasting snippets/scripts. [...]
There is already a plethora of tooling for many of these points. Not a lot of GUI stuff but ansible seems to cover a lot of ground (inventory, organized playbooks instead of shell scripts). Ansible also "just" uses SSH as a transplrt mechanism.
This feels like a solution that tries to support a flawed workflow instead a solution improving the workflow itself.
Ah, I should have clarified. I'd say ServerBuddy's probably a complementary tool alongside Ansible/Puppet/Salt etc. I have used Ansible quite a bit.
When I need to accomplish something relatively large and standard(e.g. two node MySQL installation with a replica), I'd reach for my Ansible playbooks.
However, there are certain one-off flows for which I think SB has a sweet spot. Let's say I'm an agency managing web apps for multiple clients with separate VPS(es)/bare metal servers for each client.
If I get a ticket saying something's wrong with one of their app or its dependencies(let's say MySQL), it would be very easy for me to check the container health in the Docker tab and then check the detailed logs in the Logs tab to identify the issue. During this investigation, I might also check the Ports and other tabs to check external connectivity.
Another quick one off flow I can think of is to quickly check if everything's working after running an Anisble playbook on my server(s).
It might be worth writing Ansible playbooks for these workflows if one keeps finds they're doing them over and over, but I feel that a GUI option has some merit as well.
ServerCat is very similar to this, for a GUI option. If you need features suitable to a more diverse and/or complex setup, Devolutions’ Remote Desktop Manager offers a fairly full-featured free version of their enterprise-grade software.
The former is OSS, and I’m not sure how active its development is at this point, although it’s available in the App Store for desktop and mobile. RDM is proprietary, and also offers a mobile app.
Not only ssh, it has a pluggable connection scheme to include a wide range of "how do I talk to the system," which includes AWS Session Manager, docker, kubectl, and a bunch of network appliance protocols
But I also recognize that I'm not the target audience for a GUI management app so I don't mean to pile on the "you're holding it wrong" but I do mean to draw attention to any robust solution not getting stuck in a local minima or else the user will need a separate app for each mental model of what managing "a Linux" means
A big downer on the online/remote Initiatives for learning but actually an advantage for older Unis that already have existing physical facilities for students.
This does however also have some problems similar to code interviews .
reply