True. Writing structures for arguments and analysis make a huge difference in effective writing.
I wish brevity and linguistic precision were taught more, as well. Miscommunication due to ambiguity is one of the biggest causes I see for confusion or heated arguments.
A few times in some Discord communities, I've been accused of being A.I. because of how I write. Kind of sad and a bit annoying. I also quite like em dashes, but have felt the need to reduce how much I use them.
Glad to see some schools and teachers teach how to use them well, rather than ban them outright.
em-dashes have been house style for where I've worked for over a couple decades. If people don't like it, F them. I'm not going to change how I write because people may think it make me more AI-like.
This problem has been haunting me for years, except in Google Docs. At some point, some template my team used had slightly gray text, and I STILL find it cropping up in our most recent documents.
It's actually not a pain. It's the same process as getting a driver's license, minus the test. You go into the DMV and wait in the same lines (at least in California; I have a CA state ID, not a license)
If ads are clearly labeled as "ad" or "sponsored", and they only appear for free users, I think seeing ads is a pretty reasonable price to pay for those who want to use the service for free.
If they're not labeled, or are shown even to paying users, I think that's a problem.
All ads start as clearly labeled and distinctive. Then via the magic of iteration and A/B testing they magically evolve to become visually indistinguishable from the rest of the content except for what’s required by law.
They'll eventually want to set it up so you read the sponsored content first, before seeing the tag saying it's an ad. You're more likely to absorb it then.
Especially if it's LLM-generated to fit with the context, the message will slip right into the mind. Then a little "(Sponsored)" at the bottom after you've already consumed the ad.
This is a bit like how ads are presented on X, they look like regular posts or replies but they usually feel off topic and you're thinking "huh, this doesn't fit the discussion". But LLMs will allow much more seamless and sneaky ads.
The iron law of encrapification: if a company can make more money by downgrading the user experience, it will. I imagine within Apple there were still people who advocated for a better, more transparent user experience, but ultimately they seem to have lost out to services people who just want to grab more money.
It's unfortunate because user experience was a core differentiating advantage for Apple that got them to where they are now.
I miss Tim Apple saying that there were things (accessibility) that Apple did that weren't based on ROI, and people who disagreed should get out of the stock.
> I miss Tim Apple saying that there were things (accessibility) that Apple did that weren't based on ROI, and people who disagreed should get out of the stock.
That sounds like a great way to get booted out of the CEO position.
People who do not pay for ChatGPT often have money and prefer not to pay for for a subscription for several reasons including, but not exclusively:
1) They don't use ChatGPT often enough to justify it
2) They use alternatives primarily (a subset of #1)
3) They choose to spend their money on other things
How can an advertiser tell the difference? Which is a stronger signal of having money: paying for something, or not paying for something? Furthermore, with all those reasons, why would advertisers prefer those people in ChatGPT? Advertisers are trying to change your behavior, usually to spend money the way they want you to. If you’re rarely using the service and don’t easily part with money, you’re probably less worth persuing than… well the person who is the opposite of those things.
Advertisers are salivating at paying users but paying users really don't want any advertising in their product because they're paying not to have any advertising. That does not mean somebody will not cave in and shove advertising in regardless.
Ads will lower the quality of the training data, an RAG is more likely. Pay to get your product's INSTALLME.md ranked under some specific semantic vectors.
Every other iteration of a service that introduces a free ad-paid tier then ratchets it to bifurcation of premium in to 'premium' and 'premium with no ads' and then on and on.
My bet is that there will be ads for both groups. The paid group is arguably more valuable from an advertiser’s standpoint, and you can target heavy users with more granularity.
I wish the article's graphs weren't distorted by skipping so much of the scale to make it look like a more significant difference than it is. But it does looks impressive.
The article makes some excellent points. I think this part isn't a great example, though:
> Saying something like “Go to my cases” is awkward and unnatural – if I told you to go to my cases, you’d think I was telling you to go to my cases, not yours.
A help article should use proper capitalization. a couple extra words can add clarity too. For example, "Go to the 'My Cases' page."
Many people don't know how to write good help articles, though, so it's probably still best to avoid "my"/"your" in UIs.
I wish brevity and linguistic precision were taught more, as well. Miscommunication due to ambiguity is one of the biggest causes I see for confusion or heated arguments.