An AI PC is an Oxymoron

If AI really works, you should not need a device with a mouse, a keyboard, and a screen

Ben Thompson writes,

yesterday brought a new brand: Copilot+ PCs. Yes, it’s a bit of a mouthful, but it’s a trademark Microsoft owns, and it won’t be handed out willy nilly; to qualify as a “Copilot+ PC” a computer needs distinct CPUs, GPUs, and NPUs (neural processing units) capable of >40 trillion operations per second (TOPS), and a minimum of 16 GB RAM and a 256 GB SSD.

Thompson sees Microsoft extending its franchise as the operating system for personal computers by setting and implementing standards for artificial intelligence. I do not see this is a forward-looking strategy.

Thompson ordinarily focuses on issues of technology and business. That means looking at how an enterprise plans to deal with customers and competitors.

But his description of Microsoft’s vision for an AI PC is long on technical capabilities and short on use cases. Thompson does not attempt to explain how the AI PC fits into the business or consumer world.

I think of the latest AI as primarily a revolution in the human-computer interface. The superpower of Large Language Models is their ability to facilitate communication between humans and computers. This will make the form factor consisting of a screen, a mouse, and a keyboard obsolete.

The easiest PC components for AI to replace are the mouse and the keyboard. The easier that it becomes to talk to a computer, the less you will need hand-held implements as communication tools.

The most difficult component to replace is the screen. But perhaps the attempts at creating glasses with Virtual Reality and/or Augmented Reality will pay off within a few years. If so, then the personal computer will definitely become a dinosaur.

If VR and AR glasses pan out, then the form factor of the future is a headset powered by AI. The AI could be local to the headset, or it could be in the cloud.

Until the revolution in human-computer communication enabled by LLMs is ripe, the personal computer will have an advantage in facilitating certain types of communication between humans and computers. For example, without a PC, I could not maintain this substack. I use the PC to grab quotes from various sources (such as Thompson’s blog) and to draft essays.

But suppose that within the next three years, I am using an AI-powered app to write this substack. I can tell the app to summarize substacks and other sources that I follow. I can tell the app to grab passages to quote. I can tell the app the main point I wish to make in an essay and have it compose a draft in my style.

Once an app can do those things, I no longer need a personal computer. I just need a headset that can handle AI, using the cloud if necessary. I could keep up this substack while on a walk or on a bike ride.

In the future, either AI realizes its promise or it doesn’t. If it does, then we will be accessing computer resources by using our voices, possibly with gestures. If AI doesn’t pan out, we will still need personal computers and Microsoft software, but an AI PC won’t be much of an improvement over the laptop I am using now.

I see no future scenario in which masses of people will be using an AI PC. That is what I mean when I say that an AI PC is an oxymoron.