AI wins in strategy but loses in logistics: where humans are still stronger than machines
In April 2026, the Sony Ace robot, equipped with advanced artificial intelligence, defeated professional tennis player Mia Kihara. The match was played according to all the rules of the International Table Tennis Federation (ITTF), and developers at Sony have already called the event historic. A machine has reached an expert level in a real competitive sport for the first time — a true breakthrough for the entire industry.
This is not an isolated case. Looking back at recent history, AI has repeatedly proven its superiority in strictly formalized disciplines. In 1997, Deep Blue defeated Garry Kasparov in chess; in 2016, AlphaGo crushed Lee Sedol in the game of Go, overcoming the barrier of an astronomical number of moves. In 2017, Libratus won over $1.7 million in poker by learning to bluff under conditions of incomplete information. And in 2019, OpenAI Five defeated the world champions of Dota 2 live on air. All these victories laid the foundation for the current AI technology boom.
However, despite impressive successes in games and sports, machines cannot boast total dominance. In May 2026, the humanoid robot F.03 from Figure AI lost to an ordinary intern named Aime in a package sorting competition. The showdown 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. The cycle repeated continuously.
The result: Aime processed 12,924 packages, while the robot processed 12,732. This means the human spent 2.79 seconds per item, and the machine spent 2.83 seconds. Notably, the employee had breaks for rest and lunch as required by California law, and the AI only pulled ahead in the fifth hour while the human was away. To be fair, by the end of the experiment, the intern had developed blisters and a very tired hand. The robot, however, can work non-stop, so the human's minimal lead over a short distance does not guarantee long-term efficiency.
Furthermore, there is an important economic argument. Today, employers widely acknowledge that hiring people is more profitable than maintaining AI. Corporate spending on technology is growing too fast. Microsoft is limiting internal licenses for Claude Code for staff due to token costs, and Uber exhausted its entire AI budget for 2026 in four months. Per-minute payment for computing power often eats up all the benefits of workforce optimization.
The conclusion suggests itself: machines win where algorithms are clearly measurable, but in physical work and financial costs, humans still hold the lead.
My expertise: The AI market is overheated with expectations, and this experiment is a stark reminder that the human factor and cost economics remain key barriers to total automation. Investors should take a closer look at companies that balance AI implementation with the real cost of operations.