Alexa often asks me, unprompted, if I’d like it to give me a heads-up when snow is forecasted. It’s an unexpectedly proactive gesture. Except here in the San Francisco Bay Area, snow is never in the offing—and Amazon doesn’t seem to have programmed its voice assistant to understand the command “Alexa, stop asking me that pointless question over and over.”

I don’t mean to blame an entire class of tech products for one offender’s sins. Still, Alexa’s dunderheadedness feels emblematic of the state of old-school AI voice assistants, a field that also includes Apple’s Siri and Google’s Assistant. Way back in 2011, when Siri debuted on the iPhone 4s—the day before Steve Jobs died—it felt like the future of human-computer interaction. Within just a few years, however, the consensus among Apple fans shifted from amazement to disappointment. Alexa and Google Assistant are also widely regarded to have fallen short of their potential.

When ChatGPT came along, the old voice assistants really began to look dense. OpenAI’s bot wasn’t a direct competitor, and much of the stuff assistants could do—from setting reminders to controlling smart-home gadgetry—was beyond its skill set. But the sheer fluidity of its chatter and sprawling (if imperfect) scope of its capabilities took artificial intelligence to places Siri, Alexa, and Google Assistant had never gone.

I bring all this up now because there’s a decent chance some transformative AI assistant improvements could be announced in the next few weeks, as Big Tech companies hold their developer conferences. For instance, at Google’s I/O keynote on May 14, we might hear about the future of Google Assistant—or non-future, if the Gemini chatbot continues to encroach on its territory. Microsoft’s Build is the following week; two weeks after that is Apple’s WWDC. If all of these events pass without meaningful news on the assistant front, it’ll be a sign that the whole category is moribund.

(Yes, Microsoft already killed its Siri-esque Cortana for Windows, but its Copilot covers some of the same ground and is a company-wide priority. Meanwhile, Amazon doesn’t hold a single annual event to showcase Alexa, so it’s tough to predict when news might come. But some of us are itching to hear from the company’s newish head of devices and services—former Microsoft hardware honcho Panos Panay.)

In their current state, the venerable AI assistants have several fundamental weaknesses. Rather than figuring out what we want no matter how we express it, they parse our requests in a far more rudimentary fashion than ChatGPT and its rivals. They’re hard-coded to perform specific tasks, but it’s not always obvious what those tasks might be. Despite having access to essential information about us—from the appointments on our calendars to the smart devices in our homes—they don’t always leverage it to make our lives easier.

To be fair, dragging these aging pieces of software into the generative AI age is harder than it might seem. Newer chatbots can get away with feeling a tad sluggish, but voice assistants must operate in something like real time. For the sake of reliability and privacy as well as speed, they have an incentive to do at least some of their processing right on our devices—a bigger challenge than off-loading all the heavy lifting to the cloud. If they fumble their answers or engage in outright hallucinations, they’ll quickly lose our trust.

Even so, Apple, Google, Amazon, and Microsoft have a remarkable AI opportunity in front of them. Since getting addicted to my Apple Watch Series 9, I’ve already developed a soft spot for Siri, which—in its current, somewhat creaky form—is often handy to have right on my wrist. Putting a smarter Siri there could render newfangled AI gizmos such as Humane’s Ai Pin and the Rabbit R1 superfluous before they ever gain traction in the first place.

Apple’s secretive nature makes it particularly tough to divine where Siri is going. Do reports that the company has lately been in talks with OpenAI about licensing GPT technology for iOS 18 mean that its plans for its AI assistant remain uncertain? I hope not, because it can’t afford to ponder its options forever.

Case in point: Last week, OpenAI deployed a new ChatGPT Plus feature called Memory. Rather than firing up every new chat session as a blank slate, the chatbot can now assemble a persistent understanding of users based on what they tell it about themselves. (For example, I just informed it that I love bossa nova music and hate peas, which should influence the playlist recommendations and recipes I get henceforth.) It’s a tantalizing glimpse of how chatbots can provide responses that grow more personal over time, in ways that first-generation voice assistants never have. OpenAI—which is reportedly working with Apple design legend Jony Ive on an AI hardware project—surely isn’t going to stop there.

Already, I can program the Action Button on my iPhone 15 Pro to launch ChatGPT in voice form, giving me the kind of instantaneous access that was once the domain of built-in assistants. The better that bot gets, the less relevant Siri might feel. But it would be nice to think that Apple’s assistant—a product that first intrigued me before it was an Apple product or even involved voice technology—still has some fight left in it.

You’ve been reading Plugged In, Fast Company’s weekly tech newsletter from me, global technology editor Harry McCracken. If a friend or colleague forwarded this edition to you—or you’re reading it on FastCompany.com—you can check out previous issues and sign up to get it yourself every Wednesday morning. I love hearing from you: Ping me at hmccracken@fastcompany.com with your feedback and ideas for future newsletters.