Perplexity has become a steady co-worker for me, mainly on the ‘How do I…’ type of queries (I can definitely see the role of family nerd disappearing in the future). When writing, I use for research and honing in on things, I find it hits limits fairly quickly, but usually it has unearthed some interesting links to follow through on. I’ve avoided any writing help but the point about proofreading above may be useful.
OS updates every year are filled with features I never go near, a lot of Apple Intelligence too it seems. But I think their approach, AI gradually seeping into all and sundry, is probably going to work in the end.
On the other side of the coin, I’m currently restoring an oldish (1966) car. I asked ChatGPT to identify the engine type based on the engine numbers cast into the block. Completely wrong answers for capacity, transmission and year - although it seemed almost smugly confident.
Point is, I’ve had ChatGPT return me enough wrong answers that I’m now skeptical of anything it tells me. Conceptually I’m a big fan and I do use it - albeit via a browser - but it’s of little value if it’s incorrect. Hopefully by the time Apple integrates it well enough for people to adopt using it the models will have improved their error rate.
This was challenging because often “Haven’t used it” is a synonym for “No impact, it was so shit I had to turn off Apple Intelligence entirely”.
Example: Mail’s thread summaries mess up the threaded view, even if you turn off summaries—surely a bug, but no way to stop it without turning off AI systemwide. Or again, Siri’s announced notifications are always summarised, even though summaries are not otherwise wanted and aren’t displayed visually. And the Mail app’s suggested replies, as already discussed, are also impossible to disable using a preference in Mail itself. Just lots of cases where bugs and mediocrity mean that you simply can’t benefit from it, despite the potential of the features that do actually work.
So I agree with suggestion that “Negative impact” should have been an option. And until Apple make the settings more granular, sadly I’ll have to make do without the few nice features, like article summaries in Safari (which, curiously, don’t respect the global AI setting, but which do require article highlights …).
I already have SaneBox for email categories, and it’s superior to Apple’s AI; it’s probably easily implemented with rules, if I can only be bothered. And I find Perplexity and the ChatGPT app’s voice modes (triggered with shortcuts) to be superior to Apple’s Siri integration, though I hope Apple improves on that because it does have the potential to supercharge Siri, if only their implementation is ever allowed to exceed their marketing.
accidentally installed the photos app cleanup plugin. would like to delete it plus its annoying button. what a worse than useless feature. yet more proof that there’s no intelligence in that artificial …
I’m afraid that’s typical ChatGPT garbage (or in newspeak: “hallucinations”). That setting has always been off here and nevertheless, ever since 15.5 I’ve been getting these inline suggestions. The only way I can turn them off is to turn off Apple Intelligence entirely.
It’s close, but incomplete. You want to turn off both “inline predictive text” and “show suggested replies”. Inline predictive text looks a lot like spelling autocorrect. When that was on, I was getting lots of weird accidental changes to my writing (I may type faster than that works), like inverted word order. “Suggested replies” is the thing that forces you to hit escape before you can hit the return key - it usually kicks in the moment you start to type a reply with suggestions to use “dear” instead of “hello”, or “hello” instead of “dear”.
And those settings are impossible to find! Why are they buried under keyboard? and then under “input sources”?
I asked Apple Intelligence a question a few months ago. The answer: Ask ChatGPT. I’ve never used Apple Intelligence since then and always use ChatGPT. It gives me a wrong answer every once in a while, but not very often. It has helped me out of a bind more times than I can count.
If I, after asking that question for the CVC, ask it “and what about OR?” it says it’s the same as CA, that is, you cannot turn right on a red arrow after coming to a complete stop. But according to what you posted per the OR DMV that is wrong.
I have to say though, IMHO the OR logic doesn’t make any sense. You’d install a right arrow instead of a solid red specifically to indicate no right on red (as in CA). If you want right on red rules, just use a regular red light (which could still be combined with a green arrow, if necessary). For once, I feel our way (the CA way) makes more sense.
For all those who have issues with ChatGPT, make sure that it’s doing a search rather than answering from its training data. In nearly every case where I’ve gotten a response that seemed really wrong, asking it to “confirm that with a search” pulled up the correct answer.
Simple answer: No.
The poll link opened in Firefox to a blank page so I can’t fill out anything.
I don’t want anything to do with Apple or anyone else’s “AI”.
My iPhone 13mini I don’t think can use it, nor MBP on Sierra but MBAir on 15.2 might. I tried to avoid downloading Apple-I and saw some notation in Settings saying A-I use required two language settings to be the same so I make sure the languages are different so the Mac can’t use it if it does somehow get installed.
I upgraded to the 16 Pro Max for 3 reasons; to play with the AI, to use the new camera features and lenses and hopefully to benefit from the richer Siri interface. Had no luck with any of the 3. The AI is not intuitive and rarely gives me a better answer than a regular search, the camera features and extra side button are complicated and difficult to use quickly and Siri is slightly more intelligent than a sleepy chihuahua if not downright wrong or confused. I am disappointed in the 16 and its underwhelming software. I’m surprised so often by garbled software (the camera options, even after a training session at Apple), even Safari results, and more. Can I go back to the iPhone 7, which gave me the best photos I’ve taken, but with one lens and did the simple things I need from a phone? I’m planning on buying a small, quick focus camera to take the place of the 16 mess. By the time I turn off the extra features, the photo is gone. I could go on. I won’t upgrade to the 17.
Thanks @ace the link opened fine in Orion. Then I checked and realized google.com is blocked in Little Snitch for Firefox but not Orion.
Filled out the poll, thanks for posting it, looking forward to the results!
I haven’t used many features, but I like priority notifications, I absolutely love notification summaries - I don’t care if they’re occasionally misleading, and I hate that apps can opt out; I think I should be able to decide that I want them - and I’ve used “hey siri, ask ChatGPT …” a few times, and I think it’s made Siri responses last more useful. Everything else so far is meh.
As a linguist, I find it interesting that Apple would essentially be seduced by the “a” of the abbreviation AI, and make massive investments and all sorts of claims about how it was going to use conventional AI in specifically Apple ways. We pretty much all must agree that this has been a cosmic failure. And I think the great hazard of this bad start is that even if they do make it better, we’ll be reluctant to give it another try, based on an initial bad experience. But I watched an interview with Apple execs on AI yesterday and they insisted that it was at work in many ways that were not readily apparent to users. So maybe it is, I don’t know.
Ultimately, I think the current trend of every big tech company rolling out a half-baked “AI” solution is going to poison the term for a long time. And I’m OK with that.
In the long term, there will be a variety of useful app features (photo editing, text summarizing, speech recognition, etc.) which people may or may not use, depending on how well those features work, and people will stop caring if the underlying software is based in “AI”, a human-designed algorithm or anything else.
As for privacy concerns, once most consumer devices get powerful enough for on-device ML processing (Apple is getting pretty close, and other platforms are not far behind), most of those concerns will fade into the background, much like what we think about other kinds of web servers tracking data. Something important, but no more or less important than all other kinds of tracking, and something that can be avoided by avoiding certain sites/apps.
I’m a developer and use chatGPT on a daily basis. I find that it gives pretty good answers to programming questions that are highly specific, with lots of detail. The more general, the more likely to get hallucinations. I have seen it improve drastically after the past 2+ years. I have also used it successfully for planning the outline of a 2000-mile road trip last fall, and for lots of small random stuff. I’m glad to be using it for the most part. I will retire at the end of the year, and I do worry about the future of my profession, and society in general, due to the unintended consequences of “AI”.
In the 80’s, people were concerned about AI in the form of expert systems. There was panic that the software would make all kinds of high-paying professions like doctors and lawyers obsolete. Until companies started shipping products and the world found out that they were not nearly as good as the vendors’ marketing departments were claiming.
Expert systems are used today, but only in certain niche areas, where they have proven particularly useful. And the rest of the world has moved on to chase the next shiny thing.
I think today’s neural-net-based ML software is going through a similar fad. Ultimately, it will prove useful in certain areas (e.g. speech recognition, real-time translation, image recognition) and people will abandon it in other areas.
The LLM chatbots that everybody is so afraid of today are going to be relegated to tech-support phone trees, replacing today’s minimum wage script-readers, and may become an integral (and no longer special) part of search engine queries.
Replacing everybody’s jobs? Not unless these bots can prove themselves capable of actually doing all those jobs without a real expert driving the process. So far, I haven’t seen much evidence of that and I don’t think we will.
And then the world will start chasing the next shiny thing (quantum computing, perhaps?) and will start claiming that it will be the thing that renders human beings obsolete.
and did the simple things I need from a phone? I’m planning on buying a small, quick focus camera to take the place of the 16 mess. By the time I turn off the extra features, the photo is gone. I could go on. I won’t upgrade to the 17.