I was just about to accept my friend’s suggestion of a restaurant for lunch – an Indian place on the 2nd – when Apple Intelligence came along with a different idea. “How about the Ritz?” appeared as a suggested answer above the keyboard, highlighted by the telltale AI rainbow light. The other suggested answer, “Sounds good!”, was much more sensible. But I ignored both, typed my affirmative answer, hopped on my bike, and rode into downtown Seattle, where, to my knowledge, there are no Ritzes.

Suggested answers aren’t new in iOS 18.2, but they’re part of the Apple Intelligence feature set that’s rolling out with this week’s public release of 18.2. These suggestions, which I got while planning lunch, kind of sum up my experiences with Apple’s AI so far: sometimes helpful, sometimes a little off-the-wall, and often good for a laugh. But once the novelty wears off, it’s easy to ignore – just like the AI ​​feature sets on every other so-called AI smartphone I’ve used this year.

Apple took its time getting here. The first AI features dropped with iOS 18.1 in late October, including notifications and email summaries, generative writing tools, and a cleanup tool to remove distractions from photos. It felt like a deeply viable product, but Apple had to get it something on the market for its iPhones “developed for Apple Intelligence”.

Now, after months of beta testing, iOS 18.2 has officially arrived with a more extensive set of updates: the Image Playground app for AI image generation, Genmoji, and a ChatGPT extension for Siri. You also get visual intelligence, although for unclear reasons only with an iPhone 16 or 16 Pro. Of course, there’s more to come, but Apple has finally shipped a set of AI features similar to those from Samsung and Google. The problem is that all these phone makers are still far from delivering the AI ​​smartphones they promised.

Siri’s big update in 18.2 is the addition of ChatGPT. It still sets timers and answers your basic questions as always, but can now send more complex requests to ChatGPT. It is optional and does not require an OpenAI account, which is very convenient. It’s still as susceptible to invention as ever, but it can serve as a helpful starting point if you need help with a complex topic.

On Android, Google’s AI-powered Gemini has become the default voice assistant, and while it lacked many basic features at launch, it’s been moving closer to feature parity ever since. Now you can set timers, play your Spotify playlists, And Brainstorm dinner ideas together. It’s all well and good if you’re not sure what to do with a pile of wilted produce, but the real test will come when these voice assistants get the ability to take action on our phones. Both Apple and Google are working on this, but so far the new AI-powered virtual assistants are just chattier versions of their former selves.

Of all the updates iOS 18.2 offers, Image Playground is probably the most noticeable. It is a standalone app with a waiting list. However, once you’re logged in, image creation tools are also unlocked elsewhere throughout the operating system. Image Playground is very similar to Google’s Pixel Studio, but has much stricter guidelines – that’s mostly a good thing. Requests to create an image of Pikachu inserting a paper clip into an electrical outlet were rejected, which is good news for Pikachu.

It’s cute and you can create images of your spouse as a chef, astronaut, or whatever. They’re kind of funny if you care about them. But it’s not immune to the pitfalls that many AI imaging tools struggle with, and the results often don’t look quite right or just plain weird. Like most AI tools, Image Playground is a bit of a guessing machine and often takes reasonably good images. But it also guesses wrong, like when its version of avocado toast features a pit or steam rising from a tomato in the background of an image of hot soup. And don’t ask him to do hands because you won’t like what you see.

Genmoji sets even stricter standards, and in my experience it gets a lot of things right. But there are some pretty obvious limitations, including the fact that they’re so tiny that it’s difficult to see much detail in them. They’re supposed to be tiny, but you can forget about that and get carried away adding a bunch of things and then everything will be barely visible in the final product. It was a decent representation of me in front of a Christmas tree, drinking from a red coffee cup, which looks good in the preview but is impossible to parse at the typical emoji size. They also don’t work well in group texts with RCS, so I can’t respond to family texts with offensive emojis, which is my main use case for this feature.

New in 18.2 is the ability to instruct the AI ​​to give a font a specific sound. You are no longer tied to just “professional” or “friendly” descriptions; You can make it sound like Mr.

But as with the image generation features, it feels like an AI tabletop game at this point. We all had a good laugh when we first got ChatGPT writing sea shanties about farting; If you haven’t, I’m sorry to report that ChatGPT is extremely good at this. Having this feature built into my phone doesn’t make it any more useful – if anything, it makes it more annoying to use with iOS’s text selection and formatting tools. I pity the idiot who bought an iPhone 16 with the promise of Apple Intelligence.

This is my biggest problem with AI on phones right now. Often it does what it’s supposed to do. But it’s rarely helpful and it doesn’t feel like it solves any real problem I had. That was my complaint about this year’s devices from Google and Samsung; Now Apple is at least in discussion. But they are all in the same position and have the same pressure to deliver something in 2025 that isn’t just a collection of fun tricks – the novelty wears off quickly.