I wish that was the case for me. Unfortunately, even when I get internal emails about AI in science events at best it’s unclear. But often it’s generative AI. At this point I’m kinda just forced to assume it means genAI unless stated otherwise.

Though I guess part of that is that most research groups will specify the kind of machine learning they use. So if it’s left as AI then it’s usually generative.

Outside of that though it is true that (as seen in OP) most news articles will just label any model as AI, diluting the term entirely. And it’s probably a neural network if it’s from a reputable research group (such as in this case: a University of Cambridge group running actual human trials).

GB News… Yuck. Though good job, looks like the BBC has almost the same article.

Either way, looking at the actual published paper, they used DIOSynVax’s tech. No mention of LLMs or AI anywhere. All I can find is “computational biology” and “self-improving algorithms”.

In other words: this is using the current genAI boom to hype up/get clicks on something almost certainly designed using well-established machine learning algorithms. Oniony headline I guess, but otherwise it’s just bad journalism in my opinion.

God I fucking hate the way the term AI is used for everything now. Usually it means an LLM or some other form of generative model in laymen’s circles, but people so happily mix it up with other tools.

The backlash is not (mainly) due to mechs. Even the people upset about the mechs are more upset at the fact that it represents mech customisation not coming or another category of gear being placed behind a paywall (or time wall if you really want to grind Difficulty 1). And of course the mech buffs being balanced out by other buffs (including to automaton small arms fire, despite the patch notes explicitly saying that was already in a good place).

A lot of what follows you’ve already mentioned, but I’ll write it out for those unfamiliar anyway.

I mean different people will highlight different things, but really what most people are upset about is one or more of:

• changes that affect things that Arrowhead doesn’t seem to realise (durable damage increases affecting already-fragile sentries and emplacements)
• all new content being locked behind warbonds (and therefore no new endgame progression without paying)
• everything being bugged on release (new mech flamethrower and shield both have issues)
• fixes being tied to warbonds relating to what needs fixing/balancing (mech durability)
• old bugs not being fixed (I’ll mention climbing here, though that’s a relatively new one)
• when fixes or balancing changes do happen they’re often balanced out by other changes (mech buff -> more durable damage)
• undocumented changes (chargers having no cooldown on turning after a charge)
• vague patchnotes (“durable damage increased”)

I deliberately mentioned only examples from the latest major patch, but you could easily do this for any other big patch. Additionally outside of patch things:

• hostility towards sections of the player base (e.g.: ignoring “hardcore” players)
• major orders feeling pointless
• creative player solutions getting quickly “fixed”
• enemies feel favoured and often get to ignore rules players are forced to follow (e.g.: they can land on buildings, we can’t because it’s too OP)

And I’m sure I’m missing things still.

Since it’s ELI5 I’ll keep it very simple. It’s not like I know the exact mechanics anyway. No guarantee of pedantic correctness. I’m sure if I get anything overly wrong then someone who wouldn’t comment otherwise will correct me (please and thank you).

Let’s start from interpolation. It’s a simple maths idea: inter for between, poles for points. Let’s say you have two points. You could draw a line between them, take the middle point of that line. You’ve now introduced a new point.

This concept is used a lot in physics or maths in general. Let’s say you are writing down the speed of a car over time. You have 1 speed value per second. But you’re interested in the speed at 23.33 seconds for some reason.

Now you have a few options:

• You could take the speeds at 23 and 24 seconds and just the same as before: draw an imaginary straight line between them, and read what speed that is at for 23.33.
• You could also look at how the speed changed from 22 to 23 instead, especially if you didn’t have the 24s time written down.
• You could look at more of the speed values and try to figure out how the car’s speed changes over time, since it’s unlikely to be linear. That gets you to more complex forms of interpolation. That’s what’s used to find a more descriptive equation of motion for objects.

That may have been a bit of a tangent, but it does get us back to frame generation. We are interpolating where each pixel is between frames. Or perhaps even saying: okay, this visual object moved from X to Y, what happened between them?

The key part is: graphics already have this information. It would be wasteful to re-render an entire scene every frame, so you just look at what needs updating and how. But that means you know what happens one frame to the next. So now you just take that information and do some simple maths to figure out the in-between step, and show that to the user as well.

Performance-wise it’s not costly. The tough calculation is the update from frame to frame. It does take a bit of time though, introducing some tiny lag in your display.

Of course the actual frame gen algorithms can take a lot more data into account, but the simple idea is: between Point X and point Y there exists a point A which we can calculate relatively cheaply and display first.

There’s a bit of a counterpoint to that: spaced/distributed learning contributes to long term memory encoding. Revisit something a week or two later and a year down the line you’ll remember it more than if you did it the next day.

So depends on your goals a bit. That said, if you can, don’t leave stuff last minute because stress is definitely not good for memory if nothing else.

I’m seeing this post a bit late, but I feel like I have to weigh in slightly, though it’s not my research area.

Note that my information extends more to academic studying, don’t know if it’s quite as true for learning more physical skills.

The main concept for learning is deeper learning. Which basically just means actually using your brain to think about the material. Things like connecting it to other ideas, pondering different implications, that sort of thing.

The reason flashcards work is because you think about what questions you could ask about the material. The reason you write by hand vs type is because it’s slower and you have to think about what’s more important or how you’d summarise the information.

I believe reading aloud typically works because it forces you to be slower and more deliberate, giving you time to actually process what you’re reading.

That said what you’ve written is helpful and mostly correct, I’m just not so certain about the framing. It could mislead some people into just rewriting notes while reading them out, for example, which is inefficient and not very helpful for learning.

A very easy-to-read source with practical tips:

• Optimizing Learning in College by Putnam et al. (2016) (Look it up on Google scholar for a free pdf)

Also as a final tip, my favourite exam prep technique: do a past paper without having looked at any notes or done any prep. Answer as much as you can just thinking about what you remember. Then go through with notes. It primes your brain for processing and storing the information.

I find it funny that we see:

Two supernaturally resilient and strong murderous humans (Michael Myers and Jason Voorhees)

Two non/metaphysical entities (Freddy Krueger and Pennywise)

… and Ghostface who is a normal human, regardless of which version.

Been a while since I played, but I didn’t even reach 1k hours and was trading stuff much earlier than that. In 850h I think I bought plat once, and made it to 21 (iirc) mastery total, with a dozen or two frames and weapons in store. The only thing I spent plat on is either intermediate currency for trading, or more slots (which you still get plenty of for free from events and such).

The game involves a lot of grinding to begin with, but you can easily farm a bit to find something to sell for plat.

Also Reddit misled you a bit. The game throws 50 and 75% off plat fairly regularly at you.

Especially about the competitive thing though: I’ve never had that thought. I’ve never seen it as competitive to begin with, even in a friendly way. And afaict, there’s nothing you could do with money that would give you too much of an advantage. Maybe skip a bit of the grind, but a lot comes from mastery levels, build synergies, and knowing how to play your frame right… And even then outside of Steel Path and other endgame-ish content you can easily nuke an entire level with minimal effort.

All that said: yeah, it’s a free to play game and it has a few features that will make you want to spend money. You can absolutely not spend a single cent and achieve whatever you want though, which is why people praise it. It just means playing the game more, so the question really becomes whether you enjoy the gameplay loop to begin with and how much time you’re willing to spend.

Just to nerd out a bit: I think this is justified. He has nanomachines, son. His eyesight is probably perfect, but he adopts a different persona most of the time. He’s a puppeteer and a politician most of the time.

This is when he gets his actual boss theme lyrics too and he finally treats you as an actual fight. No more metal gear, no more speeches. You get the treatment of an equal rival.

Or: Rising Revengeance might like symbolism, but it prefers the subtlety of a giant boulder and memes, and him taking off his glasses just fits the rule of cool.

I’ll add my own anecdote since I installed CachyOS a few weeks ago and have used it daily since. Have some experience with Linux Mint from before, but in the past few years I’ve almost exclusively used Windows.

For me, everything worked with default settings out the box, but I did see the wiki specifically mention “use btrfs if it works, if not, use…”. I even got my *arr stack and Jellyfin up and running relatively painlessly. And some games and programs not made for Arch/Linux.

The thing is I say relatively painlessly, but some of them involved a day of tinkering, diving into the Cachy and Arch wiki pages, etc. I’m fine with that, I find it fun. It’s the price you pay for wanting the benefits of the distro (performance, customisability, etc). And I was very clearly warned going into it, which TBF almost made me not go with an Arch-based distro.

So yeah, they are made a bit painful to use on purpose. Or rather, it’s a side effect of the core philosophies. It’s not for everyone, but it does cater to specific groups, and I think that is good. Kinda like how not every fediverse instance is for everyone (see also: Mastodon vs Lemmy vs Piefed)

I would still without a doubt recommend Linux Mint if someone wants an easy and painless experience after Windows. Heck, because of apt it’s even easier than Windows a lot of the time. And for the stuff that doesn’t work, it’ll happen if Linux gets more traction. Sadly we’re just not there yet.

(Though apparently the main thing out of everything I use in work and outside of it, it’s damned Xbox controllers that I have yet to get around to making function)

As someone originally from Hungary: as a kid I was taught I should walk on the left (if there is no pavement), so I could see the cars I’d be sharing the side of the road with. So technically it should all be reversed everywhere.

Also there is no consistent walking side tendency anywhere, and just walking to/through the shops you’d know that. Brits might do a queue but see if you can predict which side they’ll be walking on, and it’s no different elsewhere.