Steam is back from a rather confusing game sale by beginning its own mad science space to play with AI and machine learning, and some of the things, it’s coming up with are sort of fascinating.
For context, Steam recently held its annual summer sale, and, as always, simply providing most of its considerable inventory at a discount wasn’t enough for Valve. There’s still a gimmick to a Steam sale nowadays, some additional incentive to buy the game or otherwise spend time on the platforms. This time it was an idiotic “race” that was complicated as heck in implementation and paid out relatively subpar rewards (unless you had been one of the few who got the free games).
So to avoid something this overcomplicated again, Steam has launched something known as Steam Labs (not to be confused with streaming stalwart Streamlabs) where it may try strange, experimental things just like the sales gimmicks in a unique space.
Out of the three features it presently offers, the strangest by far is the “interactive recommender,” which purports to use “the magic of machine learning to recommend games you may like.” It sounds impressive. However, Steam type of already does this through the daily store queue, which recommends games to you based on the ubiquitous Steam tags. What units this one apart is how you can filter the games on a sliding scale of “popular” and “niche,” and by how old they are.