Okay, let’s be honest. Everyone wants to know more about AI. It could upend the economy, replace people in jobs everywhere and even send killer robots to terminate humans, depending on who you ask.
However, there is a kind of AI that consumers are beginning to hear a lot – perhaps too much – about of late. In the past year or so, they’ve not stopped being told about newfangled AI phones, AI PCs and AI gadgets of all shapes and sizes (yes, sorry, we have told that story too).
How are these new gadgets different from before and what exactly do they now do that couldn’t be done previously on non-AI models?
Those questions are usually hard to answer. Or the answers may not be “sexy” enough to make people open their wallets for a new AI gadget.
Let’s start with AI PC, which manufacturers are now banking on to reverse a post-pandemic slump in sales.
Early this year, PC makers were keen to show off their new AI-focused laptops and desktops that could take on everyday AI tasks that are expected to be common soon.
But what’s an AI PC? In a nutshell, it sports the right hardware to deliver processing power for tasks like number crunching and creating, say, images or even videos with AI tools and apps.
What’s the hardware involved? There are no firm definitions, unfortunately, even though everyone uses the term AI PC liberally.
It could refer to PCs that sport Intel’s new Core Ultra processors or AMD’s Ryzen 8000G that contain a neural processing unit (NPU) dedicated to running workloads such as AI effects for video calling or multi-tasking AI-accelerated software, explains ZDNet.
Just like a central processing unit (CPU) is the brains of a PC that manages the entire machine, and how a graphics processing unit (GPU) manages the graphics, an NPU is dedicated to the heavy lifting involved in AI workloads.
Unfortunately, there’s no clear “standard” for an AI PC, so it’s hard for consumers to visualise one. Bring in terms like NPU, and the selling point becomes even less clear.
There’s also the question of novelty. Some of the stuff that these AI PCs are meant to do aren’t exactly new. Optimising video calls, for example, has been done by GPUs for a while now.
And even the term NPU sounds familiar. That’s because smartphone makers have been touting it for several years now, since they started enhancing the photos you take on these portable devices.
Back in 2017, Chinese phone maker Huawei started talking up an NPU in its Kirin mobile processor. That chip, it said, would help recognise an object in an image much faster, as the AI task would be done “on-device”.
If that sounds like another current AI gadget buzzword, you’re right. “On-device” AI processing is a key feature now in phones. As the sales pitch goes, it does a lot of the AI tasks without going to the cloud.
This has two main benefits – it’s faster to do things right on the device and this also prevents your data (like your images or meeting videos) from going to the Net every time you want AI to go through them.
The only catch for today’s smartphones touting these AI smarts is that “on-device” AI has been around for several years now.
So, what do the new AI phones like Samsung’s Galaxy S24 do that aren’t possible with other smartphones? To be fair, it has embedded some of the most common AI tasks into the interface well, making the experience seamless.
One oft-cited example for the Galaxy S24 is the ability to draw a circle around an object in a photo and have Google automatically search for it.
Another is a live translation of a conversation that goes beyond the usual Google Translate experience – two people speaking different languages can have a live voice call, thanks to fast AI.
All said, however, it’ll take a brave person to say that these smarts will be exclusive to such “AI phones” for long. You can expect AI to be common on just about any phone in future, which you’d expect to run an AI app easily.
Okay, perhaps less quickly than a flagship model with a top-end processor, like the Galaxy S24. That said, it’s clear why it’s more difficult to sell an AI phone than, say, a flip or foldable phone – the benefits just aren’t as obvious.
It doesn’t help, of course, that the most common AI apps people associate with the technology breakthrough of a generation are chatbots that run on Web browsers.
When you ask ChatGPT to write an essay for you or Midjourney to create an image, the processing is done online by these AI companies.
Sure, that’s not going to be free for long, as these AI companies need to make money eventually, but for now, many users still think of AI as a magic black box where you can ask it to do tasks for no charge.
With that in mind, it’s no surprise that many will view AI PCs, AI phones and other AI gadgets with a bit of suspicion. Why do you need to fork out more for all the hardware and performance to run AI when it can be run on the cloud?
Device makers need to answer that question and more cogently explain the benefits of a new AI gadget. Failing that, AI PCs and AI phones are going to be hot buzzwords but a tough sell.
Bryan Ma, vice president of client devices research at consultancy firm IDC, told CNBC last week that it would take “a number of years” before developers figure out a killer use case that consumers can’t do without.
Echoing that, Ben Wood, chief analyst at CCS Insight, said many consumers have no idea what an AI smartphone is.
Cautioning that people might get inundated with AI talk without knowing what it really does, he used a phrase that could be increasingly common in the months ahead – “AI fatigue”.
“2024 will be the year we look back on and say that’s where it all started to happen,” he said, “but it could be a long time” before the benefits of game changing experiences become clear.
