Among the Google Pixel 8 and 8 Pro’s key new features are many functions and tools powered by artificial intelligence (AI). Many find the new AI features fantastic, while others wonder if some of them might go a bit too far.
A new AI function, Best Take, allows users to replace people’s faces with better versions. The feature has attracted significant attention from those in the tech industry, but it has also attracted a wider audience thanks to being heavily featured in Google’s advertisements and promotional materials for its new smartphones.
Best Take uses AI to replace frowns, closed eyes, and even people looking in the wrong direction. This is achieved using AI to analyze images, grab the “best” faces from other photos captured in a burst, and then blend them into a single image.
For example, imagine a person taking a family portrait, like a photo of parents, grandparents, and children. It is hard enough to get a good picture of one person, let alone an entire family. In one photo, someone blinked. In another, a person made a strange face. Someone else looked somewhere else during the shot. The possible ways that a group portrait can go awry are practically endless.
With Google’s Best Take feature, these problems are solved with the magic of AI. Something like Best Take is a miracle worker for many family photographers, ensuring that family photos always look perfect. For some, Best Take crosses a line that shouldn’t be crossed.
Of course, it is essential to note that Best Take is optional. If someone doesn’t want to use AI in that way, they do not need to. It is also worth pointing out that Best Take proposes alternative faces from a photo burst, and the user retains control over which faces are replaced and with what other faces.
“As an Instagram dad, there was an emotional payoff both to seeing my son and his friends looking their absolute most adorable, and a kind of power trip to choosing from a menu of faces which is the exact right one for that moment,” writes Washington Post columnist Geoffrey A. Fowler.
As Fowler acknowledges, the belief that smartphone images reflect reality has long been tenuous. “We already take fake photos,” he writes.
Fowler also points out that not only are many smartphone features already in place that can make people look better — like skin smoothing, artificial lighting, eye brightening, etc. — the process of cutting faces from one image and putting them on another is not unique to Best Take. People have been able to do that in Photoshop for decades, and it was entirely possible with analog photography, too.
“So then what makes me uneasy about face swapping arriving on phone cameras? It’s the power we’re handing over to AI over something as fundamental as our memories,” Fowler continues. “Many people — particularly women — are already rightly tired of society telling them to ‘smile more.’ Now a computer gets to help decide what faces are worth changing and what faces are worth keeping.”
Google says Best Take’s automated face suggestions have been developed with user demand and preferences in mind. Google also maintains that a result created using Best Take isn’t fake, at least not entirely, because the pieces that make up the photo are all real. This latter argument holds a lot of water and helps ensure natural results.
It also reflects how Google is positioning Best Take — it is a feature to help people record the version of a real, experienced moment in the best possible way, even if that way involves taking the optimal parts of multiple moments. Best Take can help people have a photo that reflects their memory of an experience, even if how that memory is preserved never actually happened. For some people, perhaps even many, the Best Take image not accurately showing a single slice of time is not a problem.
“We’ve all taken group photos where someone is looking away or has their eyes closed. Best Take is going to let parents of active kids breathe a sigh of relief…” writes WIRED’s Julian Chokkattu.
Chokkattu also observes that many smartphones include features that combine data from multiple images into a single frame, such as Smart HDR.
It is also worth considering that it is normal for people to take multiple family portraits or group shots in a row and then pick the “best” one. In a way, Best Take is just a natural technological advancement that extends that common practice.
That AI is involved does matter, though, at least to some users.
Brian X. Chen of The New York Times describes some of Pixel 8’s new AI-powered features as creating “faker photos.”
Chen cites Dr. Ren Ng, a computer science professor at the University of California, Berkley, who says of AI image technology, “This is a really big moment that’s going to change a lot of things about imagery. As we go boldly forth into this future, a photo is no longer a visual fact.”
At PiunikaWeb, Hillary Keverenge writes, “It may sound old school, but a photo best represents a given moment in time, be it perfect or not. With Best Take, you’re basically doing photoshop with faces of your friends, where you replace them with what the AI thinks are better-looking versions. Sure, the feature uses pre-existing photos to adjust the appearance, but it just feels unreal and not in the moment. Since the AI picks what it feels are the best versions of faces from different moments, the end product is a mixture of faces from multiple shots and moments in time, which feels a bit weird.”
Keverenge says they would prefer if advancing imaging technology focused more on ensuring that cameras shoot faster, offer better stabilization, and do a better job of capturing moments, even those with fast action, than more AI features.
They are not alone in this preference, although something like Best Take will be a great feature for many users. And, like so much of what Google has developed for its Pixel 8 smartphones, it is undoubtedly impressive.
That said, there is something that feels weird about AI replacing the faces of friends and family in a photo, even if the picture is technically better for it. It is up to any individual to determine where best to draw the line between reality and image enhancements for their photos and to decide how they want to use AI features when capturing and preserving memories. But for me, some of the most meaningful photos of loved ones I have include a crooked smile, a goofy expression, or a bit of bad timing. There is a unique charm to imperfection, and that is something that AI may never understand. Fortunately, for now at least, most AI is optional.