eXtensions - Monday 25 May 2026
By Graham K. Rogers
WWDC opens its doors in a couple of weeks and rumors abound. The the Musk/Altman trial was entertaining (who was worse?). AI may not please everyone, but the way weaknesses are being discovered by Mythos is valuable. Some students in the USA are expressing displeasure at Commencement speakers who laud an AI techno-future. Apple is to include more AI writing tools including a grammar checker. Netflix and Apple are both putting out a healthy selection with more to come
With the run up to WWDC (9 June) there is now some speculation on what we might see. I prefer to wait until Apple makes the announcements. Copying rumor sheets that may be wrong is not really original. Remember, this is a conference about software. There is obviously some crossover as new features may well need new hardware to run on, but the chances of hardware announcements at WWDC are usually slim. That does not stop speculation of course. Before the weekend I saw a report that claimed no major changes for the Apple Watch this year. After the weekend, when Gurman's newsletter went out, several sites are now predicting changes, including a major redesign for the Apple Watch Ultra William Gallagher (AppleInsider). I am not a lover of rumors unless there is firm evidence, and prefer waiting for any announcements from Apple. One of the announcements about WWDC appeared this week and was reported by Chance Miller (9to5Mac). The announcement about the keynote and the scheduling, has the words, "Coming Bright Up". Parse that if you dare.
Prior to WWDC, Apple has announced a host of new Accessibility features that are enhanced by AI. Ben Lovejoy (9to5Mac) reports on a fairly long list of changes and additions that include enabling use of the eyes as controls when using Vision Pro. As Lovejoy writes, "AI is about much more than chatbots. While Apple has been criticized for being slow to adopt AI features, accessibility features is certainly an excellent priority. It also indicates that there will be many other things on which the company is quietly working behind the scenes."
I just bought a new blood pressure checking device, so a report from Zayed Rehman (Redmond Pie) is of some interest. He writes that Apple may be working on a new blood pressure feature for the Apple Watch. The source for this is DigiTimes and I am in two minds about the rumors that come from this site. He reports that this could be an improvement to the already existing Hypertension Alerts feature. It is reported that Apple is seeking FDA approval and this is an important point. It this does appear, it will have to be approved by the authorities in each country and Thailand does seem to lag on these features being approved. The report also suggests that one of the Holy Grails of medical technology, non-invasive blood sugar detection, is also likely to be developed.
Apple has been interested in this for several years and at one WWDC I attended, the group I was with were specifically asked to interview a developer of an (invasive) testing solution. When I spoke to Dr Richard Milani of Ochner Health Systems - again at the request of Apple - at a health-related conference here in 2016, this was one of the ideas he discussed with me in regard to how the Watch could be used. Right now it uses red light to check heart rate, and green light for blood-oxygen levels. Some wonder if a different color light could be used to check for sugar levels. I bought a testing kit later and checked my blood daily. That was inconvenient and slightly painful, although it did reveal that my blood-sugar levels were a bit high, so I tested further at a hospital: not diabetes, but worth cutting down on sugar. Imagine the situation for those who do have to test every day.
Recently we were entertained, in a way, by the litigation between Elon Musk and OpenAI. With some of the excruciating evidence-giving in the Musk/Altman trial, I guess part of the jury's decision-making process may have depended on which of the two they disliked the least. Tim Fernholz (TechCrunch) was one of several who wrote reports on this. He notes that the jury's decision concerned when Musk had filed the lawsuit. They "found that any harms that Musk may have suffered came before the deadline for filing his claims under the law." Although, because of the decision on timing, the jury did not decide on the claims, Musk Tweeted "There is no question to anyone following the case in detail that Altman & Brockman did in fact enrich themselves by stealing a charity. The only question is WHEN they did it!" Musk will appeal we are told. Nick Roberts and Dara Kerr (Guardian) also reported on this and note that Microsoft, which was accused of aiding and abetting Altman, was also found not liable in the jury's verdict. This is undeniably good timing for the IPO that Open AI has planned.
There is more information on recently reported insecurities discovered by Calif using Anthropic's Mythos (Anurag Chakwe, Cult of Mac). The details are yet to be released. Calif personnel visited Apple to share the information about the exploit. This is good for manufacturers and developers who can run such AI software to save time looking for bugs that sometimes take years to come to light. I wonder if Apple will buy or develop its own version?
All developers discover that there are bugs in their software at some point. It is best to find these before release of course, but although some problems are revealed in beta testing, many exploits are found by the hard work done by dedicated researchers who pick apart the code line by line. That takes time.
The early discovery of such problems helps everyone, except those who would exploit the bugs. Mariella Moon (Engadget) reports that Anthropic has released a report on Project Glasswing in which it claims that some 10,000 vulnerabilities have been found. As well as Apple, the report mentions problems that it discovered in software from Microsoft who are releasing patches to cover the problems. The Engadget comments suggest that Anthropic is taking a responsible approach to how it works with software companies (and perhaps the US Government).
AI might not be as popular in the USA as some people think. Ben Stockton (Tom's Hardware) reports on some major heckling from students when speakers make reference to AI. This report was in response to the reception given to former Google CEO Eric Schmidt, whom I saw speak briefly at the introduction of the iPhone. It was hard to absorb what he said at the time, but I have gone back and watched several times and he was somewhat ahead of his time. But the students have other things on their minds.
I had been aware of the way Gloria Caulfield had been booed by the students at the University of Florida, but Stockton adds the speech of music executive Scott Borchetta at Middle Tennessee State University which has been widely reported on the Internet. BoingBoing reports on this, but also includes this somewhat tone-deaf comment from Borchetta: "Deal with it. You can hear me now or you can pay me later." Not what the students want to hear. Rather than having AI stuffed down their throats, a Commencement speech should be about encouragement. BoingBoing also provides a link to the Commencement speech of Steve Wozniak - Apple co-founder - who did mention AI without being booed, giving the students reassurance and telling them "You have AI - actual intelligence."
I feel like a voice in the wilderness here when I try to explain the damage I see to long-term student skills, particularly with academic writing, and coding. Computer engineering teachers have sort of given up. They should teach about how to use AI, of course, but students must also be taught coding basics. Other branches of Engineering should also be aware of how AI can help. Writing is different. With many tutoring sessions, and with editing, I find that as the writing develops, so the students discover more about the project they are writing on. Writing is thinking.
If writers just churn it out using ChatGPT or Gemini there is no understanding, no development. And what some think of as good writing often has major problems. I discover this when I am asked to check writing, or when an article is sent back from a publisher and I am asked to suggest improvements. I would really like to put a red line through much of it and say, start again. Instead I work through applying a Band Aid approach as best I can. It is clear that many of the writers whose work I examine do not understand the words they have used.
One of the first tools I used when Apple released its early AI features was the grammar checker. I had previously seen the results of Grammarly which has been used for a while by some of the students whose writing I am asked to examine. My problem with that type of software is always that it does not know when to leave things well alone and sometimes unnecessarily replaces words with synonyms. These may not be what the original writer intended and the students I work with, whose native language is not English, may simply accept the output. That is a problem with AI writing all over.
I tried the Apple grammar "Proofread" tool the day it became available. I tested it on a paper I had just edited. I was horrified. As may be seen in the first paragraph (below) I marked several words that had been arbitrarily replaced, spoling the original flow. Of some note is the use of "facilitates" three times in the short paragraph. That is just bad writing. I also noted the replacement of Important by "crucial" (this does not mean what many think it means) and the unnecessary "mitigates" for the perfectly reasonable Reduces. Crucial is an absolute word. There are far better alternatives.
Now we are informed by Juli Clover (MacRumors) that that among the AI improvements to the next versions of Apple operating systems there is to be a "dedicated AI Grammar Checker for Writing Tools" in iOS 27 and iPadOS27 (perhaps the Mac too, I wonder) and that it will "work like Grammarly". There will be a "translucent menu" that slides up offering suggestions. There is some variation of this now on some of my devices, although it seems to work better on the Mac than on the iPad where it is often a nuisance as it hides the text I am working on.
The predictive typing that appears as I write is far better, although I sometimes miss the suggestions offered as my brain is working further ahead than the word that appears under the cursor line. She also notes that the option is expected to include grammar suggestions. I will look carefully at these features when they are announced during the Keynote, but it seems that the new features are to be optional. I am OK with that as the writer is able to make the final decision. Even now, there is unwanted input. I typed "Apple Watch" a short time ago and the non-AI Notes changed this to "Apple Watchtower". Nice idea, but not what I wanted.
I am also concerned with the use of AI in photography. I am dreading what Apple will do next to Photos. I use digital cameras, but most of the time prefer to carry a medium format film camera. I do minimal editing and avoid AI tools as much as possible. For example, if I need to clean an image (dust, fibres), rather than use Clean Up I will switch from Apple's Photos to Photomator.
|
|
One of the most well-known photographers was Ansell Adams, whose work still inspires others, including me. Perhaps one of his most famous shots was Moonrise, Hernandez, New Mexico. I saw on Facebook this week a note from the copyright holders, The Ansell Adams Copyright Trust, that they had been informed about an AI colorized version of this image that was offered for sale at a photography show held at the Danziger Gallery. As the Trust did not authorize this they contacted the gallery owner and asked for the work to be removed. There has been no response.
Comments on the posting all share a feeling of annoyance or disgust: color, AI. Local photographer, Lee Craker posted comments on his own Facebook page that link to the original posting. Adams was famous for his human input, marking up the negatives for printing (some parts more exposure, some less), that came from experience, not a computer. That the images have always been black and white, adds to their effect. Colorizing may be an interesting experiment, but this image looks odd (especially the sky), particularly when compared to the original image which is shown on several sites. A note on the Wikipedia page expresses some doubt as to whether the copyright was renewed. The exploitation of the photographer's name - using that as a selling point - is a different matter.
I later found an article by Cheyenne MacDonald (Engadget) on this who notes that the Trust did not take issue with the AI used. Instead "the exhibitor allegedly just straight up ripped off the artist's work to make money off of it", adding the point about "the nonconsensual use of an artist's name and work for commercial purposes".
Apart from science-fiction, one of my favorite movie and TV genres is crime drama. I am enjoying the week-by-week drip feed of Apple's Criminal Record (Series 2). Netflix drops the whole season at one go, so viewers are able to binge watch. I tried that with Breaking Bad but I learned to impose limits on my viewing time. They just released a new series, simply called
Legends - the persona an undercover agent develops. It is based on fictional accounts of the investigations of the UK Customs & Excise service into drug imports in the 1990s. Actors include Tom Burke who has appeared in many series and movies, and the versatile Steve Coogan. There are several worthy performances from the rest of the cast. It was written by Neil Forsyth. The series is based on a novel, although there were several undercover operations run by Customs. Reviews are almost all enthusiastic.
I had worked with Customs officers a number of times when I was in the police force in the 1970s, mainly on checking the illegal use of untaxed diesel fuel in trucks. My job was to stop the trucks, then the Customs officers took over. When they found any illicit use - paraffin, or red diesel that is untaxed for agricultural use - their powers allowed them to seize the vehicle (carrying uncustomed goods). They used that for leverage and negotiated a cash settlement, saving the time of going to court. Drug seizures would be something else.
The series was compelling viewing for me, although I was unsure about a police escort that was stopped by the criminal gang. I had escorted prison vans with high risk criminals inside on a number of occasions. I drove, and beside me was another officer, usually with a Smith & Wesson .38 that was authorised for use if there were any problems. The escort in the series did not match my experience. There was a nice touch when I heard a word used in Liverpool that is used to refer to the police: the bizzies. I had first heard this from a couple of teenage witnesses to a car theft arrest I had made on the M1 Motorway. They were from Liverpool and were known to the police there, but they were helpful to me and my colleague. I doubt they had met county police from the southern part of England before.
We should be seeing the third part of Silo soon on Apple TV. With the previous mayor apparently burnt to a crisp (speculation), and Juliette Nichols surviving, her status as hero now sees her elected as mayor. There is also a side story that was alluded to at the end of Series 2: we discover why the silos were set up, and the subsequent Armageddon. I am also looking forward to the second part of Three Body Problem which is marked on Netflix with "Another season is coming".
Apple will have a new series later this year starring Matthew Mconnaughy and Woody Harellson (Ryan Christoffel, 9to5Mac). McConaughey was good in the movie The Lost Bus, but this is a comedy series. Although the first series had been planned with ten episodes, there was a problem in the shooting and this is now expected to be reduced to eight. Matthew Mconnaughy and Woody Harellson were superb together in the first season of the True Detective series.
I have yet to watch the AppleTV series Widow's Bay with Matthew Rhys who was so good in the Netflix series The Beast in Me with Claire Danes. I am saving that for a longer, binge watch.
Graham K. Rogers teaches at the Faculty of Engineering, Mahidol University in Thailand. He wrote in the Bangkok Post, Database supplement on IT subjects. For the last seven years of Database he wrote a column on Apple and Macs. After 3 years writing a column in the Life supplement, he is now no longer associated with the Bangkok Post. He can be followed on X (@extensions_th). The RSS feed for the articles is http://www.extensions.in.th/ext_link.xml - copy and paste into your feed reader. No AI was used in writing this item.
For further information, e-mail to
Back to eXtensions
Back to Home Page