7MMO
Moderator
Anyone who's had a console knows that on a long enough timeline they will have some kind of technical bug that you need to troubleshoot, and Sony is trying to get around that reality with an AI tech support agent.
Sony has filed a patent for an AI that would keep an eye on your system, looking for any kind of red flags and then uses an LLM to write up a patch for your system, to either preemptively fix the problem or to send a report to engineers at Sony. The problem is, that unless you can turn this feature off, all the data processed on your PlayStation would potentially be visible to this agent, because to successfully monitor a system for any and all hardware and software issues, it'll likely need total access to the system.
It's also possible that the system could make small issues into much larger ones. As Patentlyze, which found the patent says "an automated system sending repair instructions to a device in your living room needs careful guardrails so a small hiccup does not become a larger one." After all, the PlayStation 5, and likely the PlayStation 6 are just specialized computers, and there have been plenty of times that I've broken something by fixing a small problem that I probably could have just ignored (or wasn't a real problem to begin with.)
There's also a possibility that, unless the LLM has particularly strong training and guardrails, that it could make a mistake when coding these on-demand patches, and accidentally brick people's PlayStations.
It's not clear when or if this patent will actually materialize into an AI tech support agent for your Playstation. However, with the death of discs for the PlayStation 6, it's likely that Sony's next generation console is going to require an always-on Internet connection to begin with. And with that constant communication with Sony's servers, an AI system like this would be trivial to implement. So if anything, I'd guess that this feature will make its debut with Sony's next-generation platform, which will hopefully give the company enough time to work out the logistics and privacy risks that are inherent to remote AI monitoring.
Continue reading...
