Don't Buy a New Phone Just for AI. At Least Not Yet

I've been reviewing phones for more than a decade. But in 2024, I found myself doing all sorts of things I don't usually do when testing mobile devices. That includes coming up with ideas for pictures that don't exist yet, calling someone who speaks another language and having AI translate our conversation in real time, and using holograph-like digital representations of my friends to line up a group photo even when they're not sitting there. 

I'm describing features found on Google's Pixel 9 series and Samsung's Galaxy phones, which are the result of an aggressive push by both companies to incorporate AI into their biggest products. Not to be left out, Apple is launching Apple Intelligence in beta this fall, a suite of its own AI-powered tools for everything from creating custom emojis to summarizing text messages. 

The scenarios above may sound futuristic and exciting, and in some ways they are. However, that doesn't mean they're worth upgrading your phone for. With their new AI features, Samsung and Google present novel solutions to seemingly common frustrations, such as not wanting to hand your phone to someone to take a group photo or not being able to speak a different language. But in practice, these tools don't feel as natural as they should, and they can create some awkwardness.

Read more: Why Gemini Isn't Just Google Assistant 2.0

Still, Google and Samsung's AI features feel like an early glimpse at where things are headed, which makes me hopeful and curious. Right now, the most significant new AI tricks on the Pixel 9 and Samsung's Galaxy devices feel siloed and built for specific circumstances, like editing a photo or transcribing a phone call. It's convenient, but I'm waiting for the day when AI can make my entire phone smarter and more intuitive. There's a real opportunity for virtual assistants and generative AI to make smartphone software more efficient, reducing the amount of time spent digging through apps and menus.

That mission is already in full swing, as evidenced by Google's Gemini helper, Circle to Search, Apple's upgraded version of Siri and new (but seriously flawed) products like the Rabbit R1 and Humane AI Pin.

That vision isn't a reality yet. A lot needs to happen before AI becomes the reason to upgrade your phone, in terms of feature development, usefulness, accuracy and safety. 

Today's AI features don't feel essential yet

Samsung, Google and Apple have all introduced (or are planning to launch) new AI features for tasks like editing photos, creating images from scratch, summarizing or rewriting text, and more. They show potential, but they don't bring essential changes to the way you use your phone that make it feel dramatically different. 

Take the newly launched Pixel 9, for example. Its headlining AI features include the ability to search for content in your screenshots, a tool for changing an object in a photo into something different, phone call summaries, weather summaries and a new app for generating images based on a prompt.

On their own, these tools are only moderately useful for specific niche circumstances. In my review, for example, I could think of only a few particular scenarios in which Pixel Studio would be useful, despite being fun to play around with. Unless you use your phone to capture photos for a project and need advanced editing tools, or perhaps you just want to have a little fun and create silly photos to share with friends and family, I can't see Reimagine (i.e. the editing option for changing objects in photos) being that useful on a regular basis either.

Read more: Samsung and Google's Mysterious Mixed Reality Plans: What We Know So Far

The other features mentioned above, like screenshot search, call and weather summaries, do seem practical. But they're also too minor to really factor into the phone-buying decision. I felt the same about many of Samsung's Galaxy AI features after using them on the Galaxy S24 Ultra and Galaxy Z Fold 6 over the last few months. 

Like Google, photo editing and content creation is a major area of focus for Samsung, as evidenced by features like Generative Edit (for removing or altering objects in a photo) and Sketch to image (creates polished images based on rough sketches). Samsung also has a handful of translation features baked into its Galaxy phones as part of Galaxy AI, such as Live Translate, which has Samsung's AI act as an intermediary between yourself and the person you're speaking with to translate phone calls. Samsung's phones can also translate entire text message threads into a different language. But again, unless you travel frequently or are regularly in contact with someone who speaks another language, it's not a reason to buy a phone.

Apple Intelligence, which is available only as part of the company's developer beta program right now, includes similar features, like tools for creating images and summarizing or rewriting text. But not all of its features are active yet, and I'll have to spend more time with it to tell how useful it is.

AI features are missing the human element

But perhaps my bigger criticism is that even when some of these tools do come in handy, they don't feel quite… human. The suggested texts I created using Samsung's Composer feature, which drafts messages for you based on a prompt, sounded impersonal and generic, which makes sense considering they were written by AI. 

Google's Add Me feature is another example of a tool that seems useful but just felt socially weird to use. I felt like I was orchestrating a photo shoot when trying to take a group photo using Add Me rather than just taking a casual photo with friends. For those who are unfamiliar with the new feature, using Add Me requires a good amount of explaining.

The resulting photos also don't always look natural. As I mentioned in my review, my coworker and I tried to create a photo in which she was leaning on my shoulder. But she ended up standing too far away, and we didn't look like we even knew we were standing next to each other in the final photo.

I felt the same way about Samsung's language translation features, as well-intentioned as they are. When using Live Translate, it can be difficult to know when to start and stop talking so that the AI translator can do its job. 

Conversation mode, which shows language translations on both the outer and inner screens on Samsung's foldable phones, also sounds better in theory than it works in practice based on my experience. The idea is to make it so that both people in the conversation can see translations of what the other is saying simultaneously by using the foldable phones' dual screens. 

But when I tried this at Le Marché des Enfants Rouges, a bustling food market in Paris, it was hard to focus on the conversation because I was mostly paying attention to what was on the screen. I ended up placing my order the good old-fashioned way: By pointing, gesturing and using the little bit of French I actually knew. CNET's Patrick Holland also tried this feature in Paris to order a coffee, and the slightly inaccurate translation ended up making the French barista feel self-conscious. 

I give Samsung and Google credit for trying to solve actual problems that some people may encounter. But these are features that I find myself going out of my way to try rather than having them fit naturally into my life.

Read more: How Samsung Decides Whether Futuristic Concepts Become Real Products

There's potential, but also big obstacles to overcome

That doesn't mean there aren't opportunities for AI to upgrade our phones in meaningful ways. What will make AI truly useful is its ability to understand the data on our phones and serve up what we're looking for with little effort. Google's Circle to Search, which lets you search for almost anything on Google by drawing a circle around it, is a step in that direction. 

The industry is moving quickly to make that happen, and the recent upgrades to Google's Gemini assistant are proof. For example, you can pull up Gemini and ask it to analyze content on your screen, such as asking questions about a YouTube video you're watching. Luckily, features like Circle to Search and Gemini may not even require an upgrade, depending on which phone you have. Circle to Search is available on a variety of Galaxy and Pixel Android phones from the last few years, and Gemini is available across Android as an app and virtual assistant and within the Google app on iOS.

Apple is also seemingly focused on this goal with Apple Intelligence, which includes a new version of Siri that can connect the dots by drawing on your personal context when answering questions. All of these changes feel closer to turning our phones into true virtual assistants that don't require as much tapping and swiping to get things done. 

But there are some really big hurdles to overcome in order to get there. Large language models, which many generative AI features are based on, still have a tendency to hallucinate, as I realized when Gemini gave me incorrect answers while I was testing the Pixel 9. And remember when the AI Overviews built into Google's search engine suggested putting glue on pizza? 

As Digital Trends and The Verge pointed out, tools like Pixel Studio and Reimagine (i.e. the feature for changing objects in photos into something different), can be used to create disturbing images despite Google's guardrails. And as The Verge notes, that's significant because these features and apps are built right into Google's phones and don't require technical know-how or special software to use them.

In a statement to CNET, a Google spokesperson said Pixel Studio and Magic Editor are meant for creativity and are designed to "respect the intent of user prompts," meaning they may create "content that may offend when instructed by the user to do so." 

"That said, it's not anything goes," the statement said. "We have clear policies and Terms of Service on what kinds of content we allow and don't allow, and build guardrails to prevent abuse. At times, some prompts can challenge these tools' guardrails and we remain committed to continually enhancing and refining the safeguards we have in place." 

With Samsung and Google flaunting their new AI features, and Apple Intelligence requiring a newer iPhone in order to work, you may find yourself wondering whether AI is a reason to upgrade. 

The answer is no, at least not yet. 

The biggest thing I've learned after testing new AI-equipped phones is that these features -- even the useful ones like Circle to Search -- require you to think about using your phone in a new way. And while the technology is moving fast, our willingness to adopt them is what's going to make them feel essential.