AI vs. Human Intelligence is a fair matchup or not ?


Should we stop thinking AI vs. Human, Think AI With Human ?

Mostly experts are having of the view that artificial intelligence (AI) is completely an automated process without any human intervention, but in reality most of the input gained by AI based systems are from humans. Concern, that AI will replace human beings in the digital workplace, is more closer to reality and it is also likely that humans and machines will work together.

This time under the subject " ai vs human " another test is carried out by DeepMind on its artificial intelligence agents that were trained to play the Blizzard Entertainment game StarCraft II.

Test environment was Google-owned AI lab that is more sophisticated one and software, still called AlphaStar, is now grandmaster level in this real-time strategy game.It is capable of beating 99.8 percent of all human players in competition.

Starcraft AI is already among the best players.

Deepmind's AI Alphastar started small. Meanwhile she plays in the Grandmaster Ladder against the best human Starcraft 2 players. The system dominates Terrans, Protoss and Zerg - each as its own neural network.

Deepmind's AI system Alphastar is already superior to about 99.8 percent of all active players in the real-time strategy game Starcraft. Meanwhile, it plays in the competitive ladder system at the highest level: Grandmaster. Professional players who also play for prize money in tournaments move on this level. Interesting: Alphastar now dominates all three playable races - Terrans, Protoss and Zerg - and should have developed according to the developers against all three races tactics.

For each of the three quite different fractions, the program uses a self-contained neural network with its own weights and training data. The current version of Alphastar has also incorporated some predetermined limits. For example, the AI โ€‹โ€‹does not see through the fog of war and has to explore the playing card, analyze the base of the opponent and develop strategies against it. In addition, the system may execute a maximum of about 264 actions per minute.

Since the summer of 2019 , Alphastar plays in the European ladder against human players. However, the program does not pretend to be undetected and plays with several agents in parallel. Some community members such as the Youtuber LowkoTV have probably already played against the program. Partly very unconventional tactics reveal the software. He was also struck by the fact that Alphastar agents always play exactly 50 games within a few hours at a time.

German Starcraft player Dario Wรผnsch - Liquid TLO - works with Deepmind and was involved in the training of Alphastar. "While Alphastar has excellent and precise control, it does not feel superhuman," he says. However, this software is not perfect yet. Although she can quickly play against different versions of herself and evaluate this, but there is a risk that Alphastar forgot ways of playing past matches again.

The team at Deepmind continues to work on his project. The developers even see the approach as universally applicable - for example in games other than adaptive computer opponents.

Amram David

Senior Contributor at DFI Club
Amram is a technical analyst and partner at DFI Club Research, a high-tech research and advisory firm .He has over 10 years of technical and business experience with leading high-tech companies including Huawei,Nokia,Ericsson on ICT, Semiconductor, Microelectronics Systems and embedded systems.Amram focuses on the business critical points where new technologies drive innovations.
Amram David