p-a_58213

I take your point, although this was not a post about a repository, but rather for an educational website (without any link to the code provided anywhere from what I can see).

I can indeed see some UI patterns that are indicative of vibe coding. So what?

Did something like this exist before, with the same level of interactivity? I certainly had not come across it. Should Show HN be exclusive to hand-crafted code that demonstrate software mastery? Where things are going, that would be a slippery slope.

I think we should be celebrating what is possible using this new generation of tools, and how the reduced barrier to entry will result in more creativity and experimentation. As for those who are asking for AI use disclosure, why stop there? Why not also ask for disclosure of the use of any libraries or templates that made implementing it a bit easier?

I would also like to add a personal perspective. Each academic and teacher has their own take on things, their own narrative which distinguishes them from the rest. And in most cases, this unique perspective has so far been expressed through a combination of spoken words, handouts, and slides.

Yet, when it came to interactive demonstrations and digital tools, we were at the mercy of wildly overpriced SaaS products, or dependent on TAs to implement some version of our vision. The homebrewed teaching aid that conveyed concepts exactly the way we wanted was simply out of reach, unless we were prepared to dedicate months of work, at the expense of other commitments.

This is no longer the case.

Greece experienced a horrific rail disaster a few years ago. Fast forward to today, the government released this app in the interests of transparency and to demonstrate progress towards adopting latest technologies to improve safety.

Having looked at the website and some of the sources, I think that vibe coding played a significant role in its development (there are tell-tale signs in the UI) - curious if others agree with me.

Regardless, for a country that has historically lagged in railway digitalisation, this is a major step, and I thought the HN community would also find it interesting.

....at best :( - more in certain universities.

If the gym is sufficiently simple and well-coded, achieving a simulation speed of 367,920x real-time (simulating 42 years in one hour) is plausible. The question is whether these simulated scenarios genuinely reflect 42 years of real-world driving experience and truly represent the information that a single agent has at its disposal when making driving decisions.

It really depends on the subject matter and the institution's focus (and tier). For disciplines where foundational knowledge remains relatively unchanged (say, Latin) recorded lectures could be an efficient way to disseminate information without requiring professors to repeat the same material. A "flipped classroom" would offer opportunities for more dynamic interaction and deeper understanding, and of course this would cost money.

However, as a professor myself in a rapidly evolving STEM field adjacent to AI, I update at least 20% of my course materials each year to keep pace with new developments. As it happens, about a third of the new content is derived from my research group's latest work. Recording lectures isn't a one-time effort; it would require constant updates to remain relevant (and let me tell you, if you want to get the voice-over right, it is a lot more time-consuming and soul-crushing than simply turning up in class and giving a live lecture).

The value of live lectures goes beyond just "transmitting" content. They offer real-time interaction, immediate feedback, and dynamic discussions that adapt to the students' understanding. This level of interaction devilishly difficult to replicate in recorded formats.

I would ramble on more, but I need to return to the lecture materials I am developing for this Friday on Vision-Language Models :P

Hmm - too broad a statement. Teaching load varies - it can be as low as 3 hours per week concentrated in a single term.

Recalling certain candidates that I have interviewed over the years, I'd say it is spot on!

a) You are confusing a congestion charge zone (CCZ) with an emissions charge zone (ULEZ) which specifically targets vehicles that do not comply with the latest emissions standards. These are two separate schemes, with different objectives. It is the later that was linked with Labour's by-election failures, in the very outer boroughs that have fairly poor public transport.

b) The INRIX scorecard is citywide. Assuming that they went with the conventional definition of "London", ie. whatever lies inside M25, this is an area of 1579 km2. The Congestion charge zone has an area of 21 km2, which is about 1.3% of the total.

> Also congestion based pricing strategies have never reduced traffic anywhere they've been implemented. Go ask London.

Wrong.

Source: I am a transport engineer. In London.

Also, the Greek islands will have a different timezone from the Greek mainland? This must be a joke.