ChatGPT to remove em-dashes from AI generated output if asked

Showing all posts about trends

ChatGPT to remove em-dashes from AI generated output if asked

17 November 2025

OpenAI CEO Sam Altman, posting on X:

If you tell ChatGPT not to use em-dashes in your custom instructions, it finally does what it’s supposed to do!

It finally does what it’s supposed to do.

Am I to think AI agents were not supposed to include em-dashes in their input (and here I go) — unless said use was grammatically correct — all along? Were agents initially supplied a list of punctuation marks and told they could be used as they saw fit?

In other words, quite indiscriminately?

How unfortunate to think poor training of agents in the correct use of em-dashes, and Oxford commas, has lead to the perception that any text containing them is AI generated.

Here’s hoping other AI companies follow ChatGPT’s lead, and the rest of us can resume using em-dashes, and other maligned punctuation marks, as intended.

‘AI Window’ lets Firefox users opt-in to Mozilla AI browser

15 November 2025

Elissa Welle, writing for The Verge, with some apparent clarification regarding Mozilla’s proposal to launch an AI browser:

AI Window will be one of three browsing experiences offered to Firefox users in addition to the private and classic windows.

Long time Firefox users were concerned Mozilla intended to turn the venerable (once venerable?) browser into a fully-powered AI app. It seems the browser manufacturer is attempting to assuage these fears by making it clear the AI Window feature will be opt-in, and the “classic”, AI free, version will continue to be available.

Meta to phase out Share to, Like on, Facebook social plugins

12 November 2025

Anna Washenko, writing for Engadget:

The company’s official line is that the plugins “reflect an earlier era of web development, and their usage has naturally declined as the digital landscape has evolved.” But Facebook also plays a much smaller role in the broader Meta business operation than it once did, and anecdotally, it’s less common to see sites running only integrations with a single social network.

Share to social media buttons were a feature on disassociated for a while, back in the day. It wasn’t easy to gauge exactly how many people used them, but I could see they didn’t go untouched.

I only deployed the Share option, rather than Like, as I thought the sharing of posts was of more value. I wasn’t a fan of the buttons that shipped with the plugin — way too much branding for my liking — and preferred to integrate icons I crafted myself, or, for a time, text only share links. I also had a share to (then) Twitter option.

While remnants of the early web continue to whittle away, the demise of the Facebook social plugins could hardly be seen as contributing to this “evolution” of the digital landscape. Thankfully.

Two numerals, six seven, are the dictonary.com word of 2025

11 November 2025

67 has been named word of the year by dictionary.com:

If you’re the parent of a school-aged child, you might be feeling a familiar vexation at the sight of these two formerly innocuous numerals. If you’re a member of Gen Alpha, however, maybe you’re smirking at the thought of adults once again struggling to make sense of your notoriously slippery slang. And if it’s a surprise to you that 67 (pronounced “six-seven”) is somehow newsworthy, don’t worry, because we’re all still trying to figure out exactly what it means.

I’m all for it. I use the phrase constantly. I say six, seven, then pause. I resume by adding, eight, nine, ten. That way people think it’s an anger management technique.

Blogging and anger management goes hand-in-hand after all.

I would prefer it if 67 were styled six-seven though, so that it looks like an actual word. But then again I think presenting numerals as a word is part of the point of using the term in the first place.

RELATED CONTENT

language, trends

AI information summaries eating away at Wikipedia reader base

29 October 2025

Just about every online publisher has experienced a decline in the number of people reading articles and information published on their websites. Search engines presently do such a good job of breaking down the main points of news reports, blog posts, and the like, that seekers of information are seldom reading the material at its source.

Online encyclopedia Wikipedia is no exception, and falls in visitors stand to threaten what is surely an invaluable resource, along with others such as Encyclopaedia Britannica.

What happens if we follow this shift in the way people obtain information to its absurd, yet logical, conclusion? If websites such as Wikipedia, Britannica, along with news sites, and many, many, others, are forced to close because no one visits them anymore, what is going to feed the search engine AI summaries we’ve become accustomed to?

In short, we’re going to see AI summaries eat the web, and then eat themselves. The onus here is on search engines, AKA answer engines, and whatever other services generate AI summaries, to use them more selectively, and wean information seekers off them.

Is that something anyone can see happening? No, I didn’t think so.

ChatGPT Atlas browser: the greatest thing since tabs in Firefox

27 October 2025

Tabbed browsing, was, says OpenAI CEO Sam Altman, the last significant web browser innovation. Although tabbed browsing didn’t become common place until around 2002, the idea dates back to 1994, with the arrival of InternetWorks, a browser made by BookLink Technologies. Altman seems to be suggesting browsers have barely changed since the early days of the web.

He made the remark during his introduction to ChatGPT Altas, OpenAI’s new web browser, last week. His words made people take notice, but Altman doesn’t seem to know his onions. Atlas is not a web browser, it is an AI-powered aggregator of information, which may, or may not, be accurate.

So far, Atlas is only available on MacOs, meaning I’ve not had a chance to try the innovative “browser” out, but certain aspects of its functionality either baffled or alarmed me, as I watched the OpenAI video presentation. To make use of Atlas, we are required to type out commands or prompts, in strikingly similar fashion to ChatGPT.

That’s not typically how browsers are used, but as I say, Atlas doesn’t seem like a web browser to me anyway. Of more concern is the way Atlas can, potentially, access files on the local drive of your computer, or if you allow it, the contents of your email app. AI scrapers, including no doubt OpenAI’s, have been trawling my website for years probably, but that’s content in the public domain.

AI bots going through what’s in my email app, and doing whatever with it, including training LLMs is another matter entirely. But Atlas is an AI browser, so buyer beware, this is no normal web “browser” if it is even one in the first place. If people want to use it, that’s for them to decide.

What’s more unsettling though, are regular browsers, such as Firefox, morphing into AI-browsers. Mozilla, the manufacturer of Firefox, which I have been using for over twenty years, is not, it seems, introducing a new browser line, instead it is integrating AI features into an existing product.

This is not a good move, we’re all going to end up running clones of Atlas on our devices, whether we like it or not. If Mozilla wants to make an AI-powered browser, fine, but develop a separate product, and let users decide if they want to use it. Leave the original Firefox, whose early predecessor, Phoenix, shipped with those groundbreaking tabs Altman spoke of, back in 2002, as it is.

Somehow I cannot see any of that happening. Firefox is going to become an AI browser whether we like it or not. There is, however a way to opt out of Firefox’s AI functionality, as New Zealand/Aotearoa blogger fLaMEd Fury, has detailed.

Blogs, a lot of them, millions of them, as agents for change

23 October 2025

Elizabeth Spiers, writing at Talking Points Memo:

The lesson for me, from the early blogosphere, is that quality of speech matters, too. There’s a part of me that hopes that the most toxic social media platforms will quietly implode because they’re not conducive to it, but that is wishcasting; as long as there are capitalist incentives behind them, they probably won’t. I still look for people with early blogger energy, though — people willing to make an effort to understand the world and engage in a way that isn’t a performance, or trolling, or outright grifting. Enough of them, collectively, can be agents of change.

As Spiers says, it might be possible to manipulate the CEOs of large media companies, but doing the same to a million independent publishers, may not be so easy.

Eight out of ten coffee drinkers prefer instant over drip

21 October 2025

In blind taste tests conducted on eighty-four people, American researchers discovered nearly eight in ten study participants preferred instant coffee over drip coffee. The findings have, needless to say, astonished some coffee aficionados.

But I’m not sure the news is that surprising. I tried for a while to get into drip coffee, but struggled. I’m no fan of instant either, but when it comes to coffee, I think my preference is for something with a little texture, a little froth.

Probably not real coffee to those who like drip brews though, but to each their own, of course.

RELATED CONTENT

coffee, trends

ISP customer hompages lists, the first web directories of the early web

10 October 2025

Via Jelloeater on Bluesky, Jeppe Larsen’s early memories of the web, from the late 1990’s:

I remember the ISP was called get2net and it came with both email and web hosting. The last bit was particularly exciting as get2net had a listing of all homepages made by its customers on their website, which was an absolute fantastic way to discover other HTML enthusiasts and of course contribute with my own handcrafted HTML manually uploaded via FTP. The web was a lot more personal, filled with handcrafted websites where people mostly just wrote about themselves and their hobbies.

My ISP in the late nineties also had a list of customer’s homepages (Internet Archive link). One of the earliest iterations of a web directory perhaps. I frequently perused the list, visiting each site regularly for a time. Some pages were not dissimilar to what you’d see on Geocities. Avril & Andrew’s home page (Internet Archive link), is one I clearly recall, on account of the easy to remember URL.

But it wasn’t just customers checking out each other’s websites.

At one point the splash page (remember those?) of my website featured a violin. I have no idea why now. I’d put a purple tint on it, with Photoshop, and liked the way it gleamed on the white background of my site. Anyway, there was some problem with the site and I’d had to call, on the phone, a landline no less, the ISP.

You didn’t get through to a call centre back then, you spoke to the people who owned the company. I forget their names, but I usually spoke to one of two somewhat sarcastic guys.

Having explained the issue, and being put on “hold” while whoever had taken call went to investigate, I heard him say to his colleague, “yeah, I’ve got violin guy on the phone…”. The colleague responded, saying something like, “oh, purple violin guy?” You wouldn’t see that sort of… familiarity today.

Despite the snarky attitude, I was pleased no end to be actually speaking to non-acquaintances who looked at my website. Occasionally the “webmaster”, the person who looked after the servers, would also reply — usually in the middle of the night — to some of my support emails.

Something else that would never happen today.

The ISP was taken over several times during the time I was with them, growing with each buy-out. The customer homepage list vanished, along with the two original staffers, whom I never spoke to again. I sometimes wonder what became of them, the ex-ISP startup founders, the then nocturnal webmaster, along with Avril and Andrew, and where they are now.

RELATED CONTENT

history, technology, trends, web design

Robotic self-driving vehicles a threat to gig-economy food delivery work

9 October 2025

Robocart, a US company, has been developing self-driving vehicles that have the capacity to deliver ten different customer orders in a single run. The service, which the company plans to launch in Austin, Texas, later this year, will see customers pay just three-dollars per delivery, pricing many people will find attractive.

But Chicago based cybersecurity and network infrastructure expert Nick Espinosa warns that such a service stands to eliminate the roles of many food delivery drivers (YouTube link), working on behalf of companies such as Uber Eats and Door Dash.

Earlier this year, I was hearing stories about Australian web and app developers taking on food delivery work, as AI apps are doing the work they used to, for a fraction of the cost. While many of these people will be able to re-skill and eventually find new work, what will they do in the meantime, if casual work begins drying up?

« older posts

newer posts »