Get the newsletter
Sony · AI / ML · Filed Mar 31, 2025 · USPTO published Oct 1, 2026 · Oct 1 release

Sony Patents an AI That Recommends Games and Explains Its Own Reasoning

Most recommendation systems tell you what to try next but never why. Sony is patenting a system that hands you both the suggestion and a plain-language explanation of where it came from.

A person plays a video game on a console connected to a television, displaying an application menu. Drawing from patent filing US 2026/0303910 A1.
A person plays a video game on a console connected to a television, displaying an application menu.All 7 drawings ↓
Publication number
US 2026/0303910 A1(US20260303910A1)
Applicant
Sony Interactive Entertainment Inc.
Filed · Published
Mar 31, 2025 · Oct 1, 2026
Inventors
Ellana Fortuna, Garganjot Kaur Kang, Ritagya Meharishi, Elizabeth Ruth Juenger, Katherine Egeland
Status when we published
Examiner raised objections; the company can respond
Document
20 claims
US classification
463/1
Examiner
SCHNURR, JOHN R (Art Unit 2425)

What Sony's explainable game recommendation system does

Ever looked at a "Recommended for You" row and had absolutely no idea why those titles were there? Sony is filing a patent for a system that fixes that specific frustration, at least on its gaming platform.

The idea is that your PlayStation collects information about you as a player, feeds it into an AI model, and gets back not just a game recommendation but also a sentence explaining the logic. So instead of seeing a mystery title appear, you might see something like "Because you finished every side quest in your last RPG, you might enjoy this."

That shift sounds small, but it changes how you relate to the suggestion. You can decide whether the reasoning matches how you actually think about your own gaming habits, rather than just guessing whether the algorithm has you figured out.

From the filing · CLAIM 1
… generating, by the gaming system using the machine learning model, the recommendation for the user device along with a reason for the recommendation, wherein the reason for the recommendation explains why the user is receiving the recommendation; …

Translation: The artificial intelligence creates a suggestion and explains why it chose it.

How the system ties user data to a stated reason

The patent describes a gaming system (think a PlayStation console or its connected services) that pulls together user-specific data, meaning information tied to your account and play history, and passes it to a machine learning model.

That model does two jobs at once:

  • It generates a recommendation, which could be a game, a piece of downloadable content, or another product tied to the user's device.
  • It produces a reason alongside that recommendation, a human-readable explanation of why this particular suggestion was made for this particular user.

Both the recommendation and the reason are then displayed to the user on their device. The claim is intentionally broad: the "user device" is described as integrated with the gaming system, so this could apply to a console, a mobile companion app, or a web storefront.

The core technical idea is that the model is trained to generate the explanation as part of its output, not as an afterthought. That's different from bolting a generic label onto an existing black-box recommendation engine.

From the filing · THE ABSTRACT
Techniques include receiving, by a gaming system, user specific data of a user and providing the user specific data to a machine learning model trained to determine a recommendation for a user device of the user …

Translation: The console feeds personal user information into an algorithm to figure out what game to suggest.

What this means for PlayStation discovery and trust

The frustration of opaque algorithmic recommendations is real, and it costs Sony money when users scroll past suggestions they don't trust. If a player understands why a game is appearing, they're more likely to engage with it or at least feel the platform knows them. That's a meaningful conversion problem, not just a UX nicety.

For PlayStation users, a working version of this could make the storefront feel less like a billboard and more like a friend's honest opinion. It also has an implicit benefit for Sony: when users can see the reasoning, they can correct it, which feeds better data back into the system over time.

Sony's sixth filing in the AI recommendations space we've tracked since May connects to earlier ideas like a mood-reading game patent and a feedback-driven camera patent, as the company keeps exploring how software learns from you.

Source: US 2026/0303910 A1 at the USPTO (PDF). Full patent text and figures are in the official publication.

Get our take in your Top Stories

Liked this breakdown? Add Patentlyze as a preferred source on Google, and our plain-English take shows up more often in your Top Stories the next time Sony patent news breaks.

There are more where this came from

We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.

The drawings · 7 sheets

7 drawing sheets from US 2026/0303910 A1 · click any drawing to enlarge

Gaming system and application menu., drawing sheet 1 from US 2026/0303910 A1
Sheet 1 Gaming system and application menu.
Entertainment system with recommendation module., drawing sheet 2 from US 2026/0303910 A1
Sheet 2 Entertainment system with recommendation module.
Gaming system connected to network., drawing sheet 3 from US 2026/0303910 A1
Sheet 3 Gaming system connected to network.
Training system for machine learning., drawing sheet 4 from US 2026/0303910 A1
Sheet 4 Training system for machine learning.
Personalized recommendations on a television., drawing sheet 5 from US 2026/0303910 A1
Sheet 5 Personalized recommendations on a television.
Steps for generating recommendations., drawing sheet 6 from US 2026/0303910 A1
Sheet 6 Steps for generating recommendations.
Computer system components., drawing sheet 7 from US 2026/0303910 A1
Sheet 7 Computer system components.
Patent filing page

More in AI recommendation

All 86 AI recommendation patents →

Watchlists this filing belongs to

All watchlists →
Reader comments

No comments yet

Start the discussion

Real name or a handle, either is fine. Comments are read by a person before they appear, so allow a little time. Keep it about the filing.