Meta this week unveiled AI that lets you chat with virtual likenesses of celebrities. Real talk, would you text with Facebook AI Snoop Dogg about video games or frozen yogurt shops?
Microsoft and Google pitch you all-knowing AI assistants to make sense of your emails, documents and calendars. But these chatbots are confined to what Microsoft or Google knows about you.
Technology companies want to prove to you, their employees and investors how awesome they are at the latest AI hotness. And often, it’s useful to toss out imperfect AI and let us try it for ourselves.
But in their zeal to outgun each other with AI, companies have overpromised and underdelivered in ways that risk disappointing you and poisoning the public’s faith in AI for years to come.
Worst of all, tech companies are ignoring essential questions: What do you want? And what is all this AI for, anyway?
Problem one: AI that doesn’t work very well
Last week, Amazon showed off a remodeled Alexa voice assistant that the company said incorporated AI similar to ChatGPT.
This AI glow-up Alexa repeatedly failed, as Washington Post technology columnist Geoffrey A. Fowler found.
(Amazon founder Jeff Bezos owns The Washington Post. Interim CEO Patty Stonesifer sits on Amazon’s board.)
As Geoff watched, an Amazon executive asked Alexa for advice on a museum to visit in Washington, D.C. The AI instead recommended a museum in another part of the country. It whiffed a question about wildfire smoke in San Francisco, too.
Amazon said that this new Alexa is a work in progress and we shouldn’t “read too much into an early demonstration.”
Fair. But how does it serve you to show a half-baked upgrade to Alexa?
The same week, Google updated its ChatGPT competitor called Bard with features that promised to summarize your emails and assist you in chores like travel planning.
Again, my colleague found that Bard invented emails that didn’t exist and suggested irrelevant marketing emails when asked for a summary of urgent messages. A New York Times columnist found similar problems.
A Google spokesperson said Bard’s features are a “beginning and not an overpromise.”
Or, check out this TikTok showing an AI-generated summary of a Zoom meeting that highlighted silly small talk about a sweater and “Game of Thrones.”
Many of these AI technologies are experiments. And it’s fine to be guinea pigs for unproven technology — as long as the stakes are low and we get some benefit from imperfect technology.
But a company has to properly set your expectations.
If Google says its AI can summarize email, it shouldn’t make up fictional messages. If Zoom promises AI can make us more productive, it shouldn’t be a creepy eavesdropper that drains your attention with information flotsam.
Problem two: Who is this technology for?
Let’s say the more-AI Alexa actually works. Do you want it?
Most people use Alexa for relatively simple questions or tasks, like checking the weather, asking the height of the Washington Monument or playing that Doja Cat song for the 100th time.
Maybe if Alexa were more capable, you would use it for more than that.
But tens of millions of Americans use digital voice helpers like Siri, Alexa and Google Assistant. To me, that’s a sign that dumb, limited-purpose technology is sometimes what you want.
Amazon said an upgraded Alexa will be better able to understand what people ask and handle more complex requests like playing music at a certain volume for a specific length of time.
One hiccup is that the technology behind the current Alexa tends to be good at different things than the technology behind the new Alexa.
Replacing Alexa with ChatGPT for turning on your lights could be like using a leaf blower to brush away dandruff. Or maybe you’d ask Alexa to play Taylor Swift and it would try to get you to leave your wife.
Meta didn’t give a coherent explanation for why anyone on Facebook would talk to a Tom Brady “wisecracking sports debater” chatbot. Just … why?
A Meta spokesperson said Bru, the AI Brady character, is intended for “sports discussion and debate, which is something fans participate in regularly.”
Problem three: Overpromising what AI can do
I marvel at some of the capabilities of AI and its improvements in the past year.
Microsoft executive Yusuf Mehdi told me last week about using AI image creation for mock-up images of a remodeled room in his home. He asked the AI to add mid-century modern touches to a virtual image of a bar and then round the edges.
The mock-up wasn’t perfect, he said, but it was a tangible manifestation of his imagination. That’s cool.
Those kinds of examples should be plenty to get us curious about the potential of AI in our own lives.
But companies risk our excitement when they tout AI as being more capable than it really is.
Google and Microsoft pitch their AI helpers as knowing your communications, personal habits and what’s happening in the world to act as all-powerful assistants. Not exactly.
For example, Microsoft’s AI “co-pilots” might know what’s in your Outlook emails. But they can’t seamlessly peer into your family’s WhatsApp messages, what an office worker is entering in Salesforce’s database, or access appointments in your Google Calendar.
These AI helpers can still be useful, but they don’t unite all the places where we spend our time and corral our information.
Companies should be straight with you about what AI can and cannot do.
Right now, it takes a lot of work to get anything useful out of it. AI is worse than current technology for some tasks. It has corporate restrictions that the fictional AI companion in the movie “Her” didn’t have.
AI has magical moments, but it’s not magic.
Hidden in the new iPhone 15’s upgraded cameras is an unpleasant surprise: You’re likely to fill up your phone’s storage much faster with the high-resolution photographs.
Geoff wrote about one buried setting in your iPhone that can shrink the photo size from your iPhone 15. That gives you more breathing room before you get the dreaded “storage full” alert.
If you’re okay saving photos on your device at a slightly lower resolution (really, it’s fine):
Open the Settings app on your iPhone 15 → Camera → Formats → Photo Mode → tap 12 MP.
Geoff also has other options if you want to hold onto the higher-resolution photos. Read more: