EPTOn January 25 2019 04:28 peanuts wrote:
Just got home, what's the verdict thus far?
Just got home, what's the verdict thus far?
It can beat at least some progamers, so far TLO and Mana in PvP.
StarCraft II: DeepMind Demonstration: Jan 24 - Page 16
| Forum Index > SC2 General |
|
pvsnp
7676 Posts
It can beat at least some progamers, so far TLO and Mana in PvP. | ||
|
Acrofales
Spain17834 Posts
On January 25 2019 04:27 Howard_Kao wrote: The micro is just too impossible for human being, if Serral forgets his mouse pad again there'll be no way that human can beat it. Google should add restriction on things like mouse movement/friction This is a fair point. The next step should definitely be a robot arm moving a mouse or something to introduce this lag and imprecision, which is a very real limitation in the game. | ||
|
Alethios
New Zealand2765 Posts
On January 25 2019 04:29 Musicus wrote: Yeah it's capped and on average uses less APM than pros, it also has an artificial reaction of 350ms which is also on the slow end of progames. Ah right. Neat. | ||
Nakajin
Canada8988 Posts
On January 25 2019 04:30 TheDougler wrote: Not much to say about the TLO series, like, it was okay, but TLO played really bad to be honest. This second series is the real deal though. This is awesome! And the 2 series were played a week of each others and it's much much better already. | ||
|
fLyiNgDroNe
Belgium3994 Posts
| ||
|
peanuts
United States1225 Posts
On January 25 2019 04:31 pvsnp wrote: It can beat at least some progamers, so far TLO and Mana in PvP. Can it only play PvP at present? Or did they choose that matchup to showcase its micro potential? | ||
|
Serimek
France2274 Posts
| ||
|
Xamo
Spain874 Posts
| ||
|
Nakajin
Canada8988 Posts
On January 25 2019 04:33 fLyiNgDroNe wrote: This is shit, AI's are going to figure out optimal strategies and tricks too fast and we won't see learning curve anymore He's doing the most standard build right now... a bit more probe I guess | ||
|
Alethios
New Zealand2765 Posts
On January 25 2019 04:34 Xamo wrote: AlphaStar teaching us that it is better to go to 24 probes? 2/3 on gas. | ||
|
Zaros
United Kingdom3692 Posts
On January 25 2019 04:32 Nakajin wrote: And the 2 series were played a week of each others and it's much much better already. It plays at an increased speed in practice too, 200 years of starcraft already played. So one week is like the hyperbolic time chamber for the AI. | ||
|
Majick
416 Posts
15 AlphaStar Agents Serral | ||
|
FFW_Rude
France10201 Posts
| ||
|
Tyrhanius
France947 Posts
| ||
|
Puosu
6984 Posts
So sick. | ||
|
TBO
Germany1350 Posts
On January 25 2019 04:38 FFW_Rude wrote: How did it know hallucination will go that route. it moved before seeing it right ? maybe AI figured out how to download maphacks :D | ||
|
FFW_Rude
France10201 Posts
On January 25 2019 04:38 TBO wrote: maybe AI figured out how to download maphacks :D That would be funny tbh | ||
|
Kazi25
Philippines236 Posts
| ||
|
Nakajin
Canada8988 Posts
| ||
|
aringadingding
472 Posts
 | ||
| ||
