Discover more from Horrific/Terrific
👍 Metaverse Party
ChatGPT wrote this newsletter for me | Apple have learned the importance of encryption | Never throw a ‘metaverse party’
Hello — the spelling mistakes and errors of grammar in this weeks Horrific/Terrific are just in your imagination, actually.
This week was good! Finally 👍. Reasons:
Let’s check in on ChatGPT because why not fuck it let’s just have a look
Apple finally figured out that whole crazy encryption thing, those little scamps
Something something please come to my metaverse party
💬 🤖 How to talk to a robot
I’m sure many of you have already been playing with ChatGPT a bunch already — in fact last week I asked it how we might harness the power of AI to save humanity. You know… to be funny. So it’s been like ten days since Open AI announced the release, which is more than enough time for tech commentators to writes hundreds of opinion pieces, a collection of essays, and maybe sit on a panel about the future of AI.
Here’s what I’ve been reading: Ben Thompson from Stratechery very neatly illustrated how ChatGPT works by asking it for the first ten prime numbers in two ways. First he asked it to pretend to be a Linux terminal, and asked for the numbers via a python script. Then, he simply asked in plain English, ‘what are the first ten prime numbers’. Some things to note:
The machine was right both times
With the python script, it didn’t run the python and get the answer that way… because that’s not how it works
In both cases, it got the answer by using the data it was trained on, which is text scraped from the web.
So this was just two ways to ask the same question, and get the same answer, with the same method
So, I’d say this is a pretty interesting way for non-technical people (like me) to ground our understanding of what’s going on — as usual, it’s easy to forget that a conversational language model is literally built to sound like a human, and also sound like it knows what it’s talking about. It knows nothing. It simply makes sophisticated guesses based on what’s in its corpus. It can’t even search the internet on your behalf. Earlier today a friend of mine asked it who won the world cup in 1996, and it said that it was Germany, winning 1-0 against Argentina. There was no world cup in 1996, and the machine was referring to the final in 2014.
The way ChatGPT finds answers doesn’t really make sense to me yet — and, it’s not even useful in teaching you things, because you kind of already have to know the answer to your question to even know if the results are accurate.
Then we have this piece in the AI Snake Oil substack, which does well to quell any concerns you may have about how LLMs could end up being an unreliable replacement Google search, or a way for students to write essays quickly and easily. Their argument is that students SHOULD be able to write essays quickly and easily, and that it’s about time English teachers embraced LLMs in the same way maths teachers embraced the calculator.
🍏 Jesus, Apple, keep up
This week privacy advocates are congratulating Apple for finally rolling out end-to-end encryption for iCloud back ups.
👩🏼🏫 What this means:
Before now, anything that you had in iCloud storage (photos are the big one) were just out there loose on the internet
The lack of encryption is exactly how we ended up with that celebrity nudes scandal from years back
No encryption also meant that law enforcement officials were able to access data that did not belong to them, thus using Apple as their sort of unofficial data collectors.
Surprise: those in law enforcement consider this move from Apple to be a bad idea. The FBI referred iCloud encryption as something that “hinders our ability to protect the American people from criminal acts”. Okay but… since when was Apple, or any tech company, supposed to help the FBI to protect the American people? It seems that ‘tech companies must be at the mercy of government departments’ is the default, and any deviation from that is an act of treason. Maybe the FBI should just be better at their jobs?
End note: try not to celebrate Apple — a large faceless corporation — for doing the bare minimum. They market themselves as privacy-first and yet have spent years storing user data without encryption, bending to the will of lobbyists just because it was the easier option. Well I say, fuck those guys, while typing this all out on a Macbook Air and only stopping every now and again to check my iPhone…
👵🏼 ’Hello, fellow kids’
The most hilarious story this week was of course the European Commission’s attempt to engage ‘the youth’ by throwing a €387k party in the metaverse and having only five people show up. I am absolutely baffled by every overlapping decision here:
The European Commission think that the EU is something ‘to get excited about’
They also think that that the metaverse is more than just another word for ‘multiplayer VR’.
They therefore saw fit to spend hundreds and thousands of euros on designing a deploying a virtual event space that hardly anyone cared enough about to visit
Where did that money come from? I wonder what they didn’t spend it on because of this. Did they not realise that 44 likes on their Facebook post advertising the event was probably a sign that this wasn’t going to work? Are they aware that there are open source tools out there for this kind of thing, such as Mozilla Hub, which allows you to cheaply throw together virtual event spaces?
I should probably know better by now, but I am honestly astounded at how gullible and out of touch these political institutions are. Even if pulling bizarre online stunts to generate interest in an economic community was a valid way of spending public money, a metaverse party was still not the right flavour of stunt.
Their first mistake was even using the word ‘metaverse’. It’s literally a word that has been co-opted by corporate entities (e.g. Meta) because they’re scrambling to keep anyone under the age of 25 interested in their products. The people who work at the European Commission weren’t even smart enough to see through the desperate and unappealing marketing push that Meta have been in the middle of for ages now.
I’d say that a great way to figure out what ‘young people’ are into would be to ignore corporate advertising, and perhaps ask the young people themselves. The metaverse is just a loose, unhelpful concept. Even people who work for Meta don’t care enough about it to engage with it. Have any of these people heard of VR Chat? Second Life? Minecraft?? No of course not. They’ve only heard of… you know… spunking money on tragic virtual parties.