Interesting. I actually remember this game very fondly. It's a well made mid-budget game for kids, in my opinion. Nothing groundbreaking, but charming, atmospheric and with fun gameplay. I was young when it came out, definitely part of the target audience (despite not caring about Harry Potter), but even revisiting it as an adult, I still found it to be enjoyable. Reading the Wikipedia article, I'm actually surprised by how low its review scores are.

How did your magazine cover it?

Do you remember which game it was?

You can tell from the photos it's not exactly high-res, but for the kinds of usage cases they are demonstrating (mainly an info display you place somewhere in your house), that's not really a deal breaker. It's not meant to be a digital picture frame or an ebook reader, where DPI would actually matter.

I only have very limited experience, but from what I've personally figured out, another good practice is to look into the thinking process (even though models can and do "lie" in there, so it's not really the actual thing that's happening under the hood) and catch mistakes early before the AI has wasted a bunch of time and (possibly) money on some nonsensical detour. You then have to stop the process early, add a line or two to the prompt to explicitly stop it from choosing this detour and start again. (In my case - I almost exclusively use local LLMs - it's not about money, at least not directly, but because there's only a very limited context memory due to the amount of RAM I have, so I don't want it to be filled up by stuff that's not helpful.)

It's like babysitting a toddler. The moment you look away, chances are it might try something you really don't want to happen.

Sometimes, the LLM might misunderstand your prompt entirely, because you chose imprecise language, made a typo that changes the meaning, you asked something of it that is way outside of its ballpark/training data, because of missing context or because the AI just had another dumb moment.

The old adage that computers are just as smart or dumb as the user (but much faster at it) often applies to AI as well, at least in certain use cases - in others, AI can actually help less intelligent users, but at the same time drag down smart people and erode their skills over time. It's weird what it's doing to our brains, which is why I'm very wary of using it for anything other than experimentation and fun, and avoid using it for things that are actually important to me. Many people don't have this choice though, with mandatory AI usage in many orgs still being on the rise.

That's certainly a unique sentence no other person has ever uttered before.

Publisher relations is another reason. I don't think EA appreciates this.

You can easily emulate them all these days. I know it's not the same, but often times, that's actually a good thing. It's far less of a hassle to play older titles on modern devices than to hook up an analog output era console to a modern TV and getting a good image quality out of the combination. Not to mention, performance and resolution were often much worse than you remember. Hell, some emulators even fix bugs and you can easily play ROM hacks or other mods that way - which sometimes also fix bugs, like a fan-patched version of Gran Turismo 2, which makes it actually possible to complete the game.

Not gonna happen. Car, music and other licensing makes this practically impossible. These licenses are only for a limited number of years - renewing them is often ridiculously expensive, to the point of usually making no economic sense. It's the main reason why old racing games like this one are not available digitally.

Even Rockstar was forced to re-release their older GTAs with some music tracks removed (and they had to patch it for existing owners as well). If it's too expensive for them and far higher profile titles, you know it's not going to happen for a game like Underground 2.

Do you think that this "majority response" is in any way representative of the user base on that site? Have you considered that the kind of person who responds like this is nothing but a very loud outlier?

I built my current system in early January 2025. I felt like prices were bleeding me dry back then (especially the GPU - I paid over 750€ for a slightly used 4070 Ti Super), but in retrospect, I lucked out with the timing, picked exactly the right moment for using the money I had saved up over a long period of time and am feeling extraordinarily privileged for having a really good machine right now. I recently needed a RAM upgrade anyway, which was extortionately expensive, but I felt like it's now or never. The machine wasn't built with AI as its main or even secondary purpose in mind - I only learned about Strata after the RAM upgrade.

That said, if you are using AI in e.g. a business context, Strata drastically lowers the barrier of entry for a serious local LLM. Two (more like three these days, I'd guess) grand isn't a large expense for anything but the smallest business (compared to the substantial five digit sum you'd have to spend without it), although it should be stressed that you can't really do many simultaneous sessions on a PC or server rack that just meets the minimum requirements for Strata - and it'll eat up most system resources as it's running, only leaving enough for very light other tasks, at least if you're trying to run the version of Qwen that does the most with your system.

Might convince some people to try out a local AI that runs entirely on one of their systems instead and is thus inherently private and can be uncensored. That is, if they ever hear about this being a possibility.

I consider myself an AI skeptic, but if one wants to use AI for whatever purpose (mine is curiosity - I don't use AI for anything other than experimentation), local AI is the way to go, in my opinion. You can run pretty impressive stuff on any halfway decent PC (I first started out with with hardware from 2017) - and if you have a powerful system (good CPU, 64 GB RAM, modern GPU with at least 12, better 16 GB of RAM), you can use a software like Strata to run AI models at impressive speeds that would normally require hardware that is out of reach for ordinary people.

It's kind of astonishing how late they are. Then again, it's Intel. The entire company has been a headless chicken for a while now.