I swear I'd read some kind of "futuristic" news article (written in 2015, yes, but faking to have been written in the future). Was I wrong. ;-) Awesome nonetheless.
This is really cool and after reading the article, I'm left with the desire to ask a bit more.
A game like 2048 is pretty straight forward and after you figure it out (i.e. you understand the 4 pieces of advice the AI figured out) the game feels like mostly chance (as the article states).
>Some games ended in a few minutes due to a series of unfortunate random tile spawns, while others nearly lasted 4 hours and reached scores previously thought impossible.
So is the AI's strength in this instance that it's relentless and doesn't get tired? Or does it have a few other insights on the game that most humans miss? Are there other comparisons of other AI's and where their advantage comes from? (strategy vs. endurance) Or is the AI's outcome distribution very similar to a sampling of human outcomes?
The article is pretty weak on how the AI works, but says that the 4 rules were built in as heuristics for the AI, which makes the statement in the article about how it figured them out seem weird.
The human records will shift to become about who made the AI. So in case you are afraid to no more be able to brag about having biggest, because of AI, fear not, it's like saying horses beat all human running records. It will just get forked into a new category of games.
It's really a matter of how long, in time, the race is. Someone in your link even mentions this. The 24 hour record for humans around 180 miles, whereas for horses it's about 100 miles. Pretty huge difference there.
Ha, good point, maybe the horse cannot run for 24h non-stop when that would be possible for a human. Nevertheless if a horse can cover 100 miles in ~6 hours, he still has 18 hours for the remaining 80 miles, so allowing him to rest (which is also pretty common in some human ultra-trails) I'm pretty sure the horse could still beat the human record of 180 miles in 24 hours.
Looking at the times of both horses and humans it seems that the main problem of this race is lack of professional competitors. When I see that the horse rider won by covering 35 km distance in 2 hours 40 minutes (on some years)then I think: that would be beaten by any professional marathon runner, who tend to beat that time on longer 42 km distance.
