Microsoft announced Copilot+ PCs and functionality at their Build 2024 developer conference. Once feature of Copilot+ is Recall, a system by which Windows will record what you are doing so that you can, well, recall it later.
As Microsoft said as they announced the next generation of Windows functionality:
Copilot+ PCs are the fastest, most intelligent Windows PCs ever built. With powerful new silicon capable of an incredible 40+ TOPS (trillion operations per second), all–day battery life and access to the most advanced AI models, Copilot+ PCs will enable you to do things you can’t on any other PC. Easily find and remember what you have seen in your PC with Recall, generate and refine AI images in near real-time directly on the device using Cocreator, and bridge language barriers with Live Captions, translating audio from 40+ languages into English.
Easily find and remember what I have seen with my PC? Well that’s not the functionality I was looking for. This is not really AI, it’s using methods and processing power to record what I was doing and when so that I can benefit from it. How, I do not know…..I rarely use my browser history and if I need to remember something, I note it down.
And of course – what could possibly go wrong? Let’s see….not touching it but just thinking in basic security building blocks at a high level:
- If I can see my Recall history, any person or process running as me on my PC can see it
- Microsoft products nowadays are released quickly and early with years of bugfixing to follow
- Having admin rights on the machine would suggest to me that user isolation is a dream
- I do not want this on my own PC
- I certainly do not want this on my employer-provided PC
Luckily for us, a researcher shoe-horned it onto his PC and got it working. Kevin Beamont assessed the current implementation of Recall and provided quite a report. It’s actually as bad as it could be, right now.
A summary of perhaps the key points he discovered:
- The database is actually quite light and does not appear to store raw screengrabs but rather is able to get text and context from the screen (hence not really AI but rather compute power to spare to do this)
- If you have a foothold on the PC, the data is available to you
- Since the data is small (possibly 50kb per day), the amount of data a modern drive could store about what you have been doing is mind-blowing (think Hello, my Perverted Friend emails – on steroids). Making an estimate of a high of 250kb per day, 1Gb of storage could be around 400 days of storage – that’s a great deal for very little space.
It could be a good idea but it has to be secure. It has to be absolutely bulletproof – protected from being extracted without authorisation but remain portable so that people with multiple devices can benefit and people who replace devices do not have to start from zero.
This wealth of knowledge, this fountain of information – it might be useful to some people. So why is Microsoft doing this and why now?
I think it’s clear; Microsoft sees more Windows and more services to be sold. Windows everywhere with a move from a learn-forward computing experience to a lean-back experience.
The Fall of the Keyboard
Microsoft desperately wanted to get into the handheld market (phones, tablets) but, with Windows Phone, they were too late to the party. And so, with the AI capabilities (including the processing power of the additional compute chips), the beginning of this migration towards a computing experience for the masses with touch and with speech being the methods of interaction, Microsoft wants to take consumers to a point where they can use their devices without a keyboard but reliably, quickly and enjoyably.
The phase we enter in this process is the gathering of massive amounts of information to create the basis for this. This basis might be on a per-user basis but the real value will be attaining a starting posture that avoids a massive learning curve for a new user or a new customer. So we have to expect that there will be some opt-in to contribute to the greater knowledge graph that is the basic service.
A new wave of computer usage will be born where the compute load is variable and Windows scales from casual use on an iPad-like device to more applied uses where there will be keyboards and mice, big screens and heavy workloads.
To support this, Windows on Arm is extremely important. x64 platforms are the norm today but they are typically power-hungry units with average or bad battery life. The aim for the new world of Windows has to be mobile-first with great battery life because that is where the majority of customers are. The masses. They use their phones a lot today and are likely to want a larger screen experience that can execute more tasks as a companion device.
I am not sure I want to have Recall on my device, slurping up all of my data and habits but I would be happy to see the end of a keyboard as this would be a seismic shift in modern computing. Engaging with an interface, spoken or via text input, where I state the abstract of what I want to my computer and get what I want from it is the future of computing..
