Looking at 56 matches in the standings, with 52 wins and 4 losses, a winning percentage of 93% , this is a terrifyingly bizarre set of statistics.
If you've ever played ranked games on the Korean server of League of Legends, or are simply a casual LOL player, you'll know how absurd this number is. This isn't about intentionally smurfing in Bronze or Silver games; this is a top-ten matchup from the Korean server, the battlefield globally recognized as the highest level of competitive League of Legends play.
What's even stranger is this account's schedule. Starting from January 8th, it played almost every day from noon until 2 AM, immediately queuing for the next match after each one ended, with virtually no rest. Despite playing for over 14 consecutive hours at such high intensity, its win rate actually increased , which exceeds the physiological limits of normal humans.
League of Legends players worldwide are in an uproar: "56 games in two days, is that even possible? Can a human do that?", "14 consecutive hours plus a winning streak, I don't think a human could do that.", "The amount of time he played is inhuman."
Amidst this wave of doubt, one name has been repeatedly mentioned:
Elon Musk.
Conspiracy Theory: Musk's Game AI
This is not just speculation from netizens. Just two months ago, in November 2024, Musk publicly announced on the X platform that he hoped to use Grok 5, the latest AI model developed by xAI, to challenge T1, the League of Legends world champion team, in 2026.
Musk even set two major restrictions for "fair competition":
- The AI can only see the screen through a camera, and its vision is no more than "20/20 vision of a normal person".
- Response latency and click rate must not be faster than human speed.
This sounds fair, but gamers know that the real challenge isn't hand speed, but the speed of decision-making and output.
If AI can calculate the optimal positioning, skill release timing, and team fight priority within 0.1 seconds, then even if its hand speed is the same as a human's, it will still have the advantage of overwhelming the opponent.
T1 captain Faker responded to the challenge, saying, "We will win the game, but AI will win sooner or later, just not next year."
This statement is intriguing. Faker didn't say that AI could never win; he said "sooner or later."
Now, could this mysterious account with a 93% win rate be "sooner or later" coming?
Does Musk have this capability?
Let's analyze this rationally: does Musk's xAI team really have the capability to create such AI?
1. xAI's "StarCraft DNA"
Igor Babuschkin, one of the co-founders of xAI, previously led Google DeepMind's AlphaStar project, an AI system trained through reinforcement learning that eventually defeated human players and reached master-level performance in StarCraft II.
In other words, someone on the xAI team has defeated top human professional players in the field of real-time strategy game AI.
StarCraft II is just as complex as League of Legends, and even more challenging in terms of micro-management and multi-tasking. If they could do it back then, why can't they now?
2. Musk's "Obsession with Game AI"
This isn't the first time Musk has challenged gaming AI. Back in his OpenAI days (when Musk was a sponsor of OpenAI), the organization developed AI capable of defeating professional teams in Dota 2. Musk has an almost obsessive interest in the idea of "AI challenging the strongest human teams."
When AI researchers suggested testing Grok's performance in a real-time strategy game, Musk even responded positively: "I hope to see Optimus (Tesla's humanoid robot) operating the mouse and keyboard."
This is someone with a deep obsession with "AI playing games." And when someone with a top-notch AI team, ample funding, and strong motivation sets their sights on something, it's hard to say they can't do it.
3. Korean server is the best training choice.
If you were to train an AI capable of challenging world champions, which server would you choose? The Korean server, of course! It's the server with the highest global competition intensity and the highest density of professional players.
The behavior pattern of this mysterious account matches the logic of AI training:
- Continuous, uninterrupted battles (no need for rest, meals, or sleep)
- Reschedule immediately after each session (to maximize training efficiency).
- Win rate increases over time (a typical characteristic of reinforcement learning).
- The primary position is the mid-lane (the position that tests decision-making and map control the most).
The obvious opposition: This is human nature
Of course, there is another perspective.
"They're LCK trainees, young people have a lot of energy," and "Only substitutes play ranked games all day." These arguments aren't entirely without merit. The Korean esports industry is highly professional, and many young players do indeed undergo grueling ranked training to earn playing time.
The problem is that even professional players rarely maintain a 93% win rate in the top ten of the Korean server.
Keep in mind that when a player reaches the Grandmaster rank, their opponents will be world-class players like Faker, Chovy, and ShowMaker, or their substitute trainees. At this level, maintaining a 60% win rate is already top-tier; what does 93% mean?
This means that this account is not just strong, but "stronger than Shohei Ohtani".
Another detail worth noting is that this account almost never interacts with anyone. Normal professional players or trainees will communicate with teammates and leave traces on social media during ranked matches. But this account is like a cold machine, only inputting and outputting, devoid of emotion and social interaction.
Expert Perspective: AI Has Already Proved Itself in Esports
"AI challenging esports players" is not science fiction, but a reality that has already occurred. Here are some past examples to remind you.
OpenAI Five: A Dimensional Reduction Attack on Dota 2
In 2018, OpenAI's AI team, "OpenAI Five," defeated a human professional team in Dota 2.
This AI was trained through the equivalent of 180 years of gaming time each day, and eventually mastered the team battle timing and strategic decision-making that would take humans several years to grasp.
AlphaStar: Master-level StarCraft combat power
Also in 2018-2019, DeepMind's AlphaStar reached Master level in StarCraft II and won against professional players.
The terrifying thing about this AI is that it can handle multiple battlefronts simultaneously, accurately calculate resource allocation, and predict the opponent's tactical intentions—all things that the human brain struggles to do at the same time.
Faker's warning: "AI will win sooner or later"
When Faker said "AI will win sooner or later," he wasn't joking.
As the greatest player in the history of League of Legends, Faker understands better than anyone that the core of the game is not hand speed, but decision-making.
Decision-making is precisely AI's strength.
Esports analysts generally believe that when AI can "understand" game rhythm, teamwork, and risk assessment like humans, human players will lose their final advantage. This is not a matter of hand speed, but a difference in cognitive design.
How will AI change the landscape of esports in the future?
Regardless of whether this mysterious account is ultimately an AI or a human, one thing is certain:
The question of AI entering esports is not whether it "will" or "will," but rather when.
Imagine if every professional esports team had a "never-tiring AI coach" that could simulate the style of any player in the world, test the feasibility of any tactic, and provide real-time decision-making advice. This could increase the training efficiency of human players by 10 times or even 100 times.
AI can act not only as a player but also as an analyst. By calculating team win rates in real time, predicting key decision points in the next 30 seconds, and analyzing changes in players' psychological states, viewers will gain an unprecedented "God's-eye view" viewing experience.
Perhaps future esports competitions will no longer be a battle between "humans vs. AI," but rather a collaborative competition between "humans + AI" and "humans + AI." The strongest force will not be AI, nor humans, but rather humans who best understand and master AI.
When AI can perfectly execute any action, the value of professional players will shift from fast hand speed to "decision-making ability." Those players who can understand their opponents' psychology and make creative tactical decisions will become the true MVPs.
The truth is not important.
Returning to the original question, who exactly is this mysterious account with a 93% win rate?
Perhaps the answer is no longer important.
Importantly, when we begin to lose the ability to distinguish between human and AI opponents, the rules of the game have changed forever.
If it's a human, this might be the emergence of the most amazing genius in the history of esports. If it's AI, this would be a historic moment for artificial intelligence entering the esports arena.
Either way, we are witnessing history.
For Musk, regardless of whether this account is a test of Grok 5, he has already achieved a major goal: to make the world seriously consider a question: when AI can defeat us in the areas that most require human intuition, creativity, and teamwork, what will be left of humanity's "irreplaceability"?
