Reinforcement learning is a machine learning process in which autonomous agents learn to make decisions by interacting with their environmentCITE:E1. Rather than following instructions, the agent learns to perform a task by trial and error, with no guidance from a human user during the processCITE:E2.
Reinforcement learning does not rely on labeled examples of correct or incorrect behavior, which is what separates it from supervised learningCITE:E3. Where supervised learning trains on pre-tagged data, a reinforcement learning agent instead builds its understanding through its own trial-and-error interactions with the environmentCITE:E2.
An agent's only objective is to maximize its cumulative rewards from the environmentCITE:E4. It pursues that objective through a policy — a function that maps the agent's observations of the environment to the action it will take nextCITE:E5.
AlphaGo, developed by Google DeepMind, won the first-ever match between an AI system and a Go professional by a score of 5-0 in 2015CITE:E6. It followed that result with a 4-1 victory in Seoul, South Korea, in March 2016, a match watched by more than 200 million people worldwideCITE:E7.
| Match | Date | Result | Viewers |
|---|---|---|---|
| First AI vs. Go professional match | 2015 | AlphaGo won 5-0 | Not reported |
| Seoul, South Korea match | March 2016 | AlphaGo won 4-1 | Over 200 million worldwide |
Taken together, the two results show a consistent pattern: an agent operating without labeled instruction, guided only by a policy that maps observations to actions and an objective to maximize cumulative reward, won both its first-ever professional match and its highest-profile public matchCITE:E4CITE:E5CITE:E6CITE:E7. The trial-and-error framework defined by IBM and the policy mechanism defined by Google describe, in the abstract, the same process that Google DeepMind's own results describe in practiceCITE:E1CITE:E5CITE:E7.
Reinforcement learning is a machine learning process in which autonomous agents learn to make decisions by interacting with their environmentCITE:E1.
Reinforcement learning does not rely on labeled examples of correct or incorrect behavior, which is what separates it from supervised learningCITE:E3.
An agent's only objective is to maximize its cumulative rewards from the environmentCITE:E4.
AlphaGo, developed by Google DeepMind, won the first-ever match between an AI system and a Go professional by a score of 5-0 in 2015CITE:E6.
The mechanism worth focusing on here is the policy: a function mapping observations directly to actions, paired with a single objective of maximizing cumulative reward, and no labeled examples to lean on. That framing explains why AlphaGo's two results are worth reading together rather than separately — a 5-0 sweep in the first-ever professional match and a 4-1 win in the far more scrutinized Seoul match, watched by over 200 million people, are two data points from the same trial-and-error process, not two different techniques. The indicator worth watching next is consistency: whether a policy trained this way keeps producing decisive, lopsided results as the scale and visibility of the contest increases, the way it did between 2015 and March 2016.
Google announced a €13 billion ($15.1 billion) investment in Finnish AI infrastructure to be deployed in 2027–2028, its largest single investment in Europe. The plan covers new data centers in Kajaani, Muhos, and Vaala plus an expansion in Hamina, backed by a 22-year power deal for 50% of Loviisa nuclear plant output and projected to support 37,000 construction-phase jobs and 7,000 permanent roles.
Chenbro (勤誠, 8210) reported August 2026 revenue of NT$2.457 billion, roughly flat month-on-month (-0.2%) but up 40.5% from a year earlier. Cumulative January–August revenue reached NT$19.848 billion, up 51.7% year-on-year.
Chang Hwa Bank (TWSE: 2801) reported cumulative after-tax profit of NT$15.567 billion for January through August, up 23.21% year-on-year with EPS of NT$1.29, a record for the period. The bank attributed the gain to loan, deposit, and wealth management momentum, and August alone delivered NT$2.916 billion in profit, up 29.16% year-on-year.