Human vs Machine: where AI is truly stronger, and where humans remain unrivaled
April 2026 became a landmark month for robotics: Sony's AI-powered robot Ace defeated professional tennis player Mia Kihara for the first time in an official match under International Table Tennis Federation (ITTF) rules. Sony developers call this event historic — the machine achieved an expert level in real competitive sports, marking a true breakthrough.
However, despite these impressive achievements, it is still premature to talk about complete machine dominance. Let's examine the key milestones in the human-AI rivalry and the areas where humans still hold the lead.
Five historic AI victories over humans
Over the past decades, technology has repeatedly proven its superiority in strictly defined disciplines. Here are the most striking examples:
1997 — Chess. The Deep Blue supercomputer defeats world champion Garry Kasparov. This is the first victory of a machine over a reigning champion in a classical match.
2011 — Jeopardy! quiz show. The IBM Watson system crushes top players, demonstrating the ability to recognize complex speech patterns and extract knowledge from vast data sets.
2016 — Go. DeepMind's AlphaGo program defeats Lee Sedol with a score of 4:1. Go is considered a game with an astronomical number of possible moves, making this victory particularly significant.
2017 — Poker. The AI program Libratus wins over $1.7 million in chips against professional players. Here, AI learned to bluff and make decisions under conditions of incomplete information.
2019 — Esports (Dota 2). The OpenAI Five system defeats world champions, team OG, live on air. This required coordinating the actions of five agents in real time.
These five breakthroughs, which became well-known long before the AI technology boom, laid the foundation for the modern industry.
Humans still win: physical labor and economics
However, machine dominance is not an absolute rule. In May 2026, Figure AI's humanoid robot F.03 lost to an ordinary intern named Aime in package sorting. The contest lasted 10 hours and was broadcast live. Each participant had to scan a barcode, lift a box, and place it label-side down on a conveyor belt.
In the end, Aime processed 12,924 packages, while the robot handled 12,732 units. This means the human spent 2.79 seconds per item, and the robot 2.83 seconds. Notably, the employee had breaks for rest and lunch under California law, while the AI only pulled ahead in the fifth hour when the human stepped away. By the end of the experiment, the intern had blisters and a very tired arm. The robot, however, can work non-stop, so the human's minimal lead over a short distance does not guarantee long-term efficiency. Currently, physical labor allows humans to stay ahead, but for office workers, the situation may change faster.
Additionally, there is an important economic argument. Today, employers widely acknowledge that hiring humans is more cost-effective than maintaining AI. Corporate spending on technology is growing too rapidly. Microsoft is limiting internal licenses for Claude Code among staff due to token costs, and Uber exhausted its entire AI budget for 2026 within four months. Per-minute computing fees often eat up all the benefits from workforce optimization.
My expert opinion: Machines win where algorithms are clearly measurable — in games and sports with fixed rules. However, in physical labor and financial costs, humans still hold the lead. And until computing costs drop dramatically, human labor will remain economically more attractive for most tasks.