AI Development: Where Does the Journey Lead Next?
In October 2017, researchers at DeepMind, led by David Silver and his team, unveiled a groundbreaking AI called AlphaGo Zero. This innovative program, a significant advancement in the field of artificial intelligence, not only outperformed its predecessor but also redefined the way we perceive AI learning.
From the very first day of its training, AlphaGo Zero demonstrated remarkable prowess, playing at the level of an advanced professional Go player. After just two days, it surpassed the performance of the version of AlphaGo that famously beat Lee Sedol in 2016. By the end of its 40-day training period, AlphaGo Zero's Elo rating was an impressive 5,000, far ahead of Ke Jie's rating.
Unlike its predecessor, AlphaGo Zero avoids the limitations of supervised learning. It starts only with the rules of the game and a reward function, teaching itself the game without recourse to human experts. This self-taught approach could potentially benefit any problem that boils down to an intelligent search through an enormous number of possibilities.
AlphaGo Zero learned to play Go by studying thousands of games between expert human opponents and refining its strategies through millions of matches against itself. It even discovered and preferred joseki sequences that were entirely of its own invention.
The original AlphaGo, on the other hand, used a method called supervised learning, which relies on studying examples of human games to understand tactics and strategy. This method is useful for many AI tasks, such as face recognition, speech recognition, spam filtering, and so on.
The victory of AlphaGo Zero over the version of AlphaGo that first beat Ke Jie was decisive, with a score of 100 games to zero. This demonstrated the power of machine learning in artificial intelligence and encouraged experimentation by repeatedly playing games against other versions of itself, subject only to the constraint of maximizing its reward.
Advances in AI, such as AlphaGo, could potentially help push human players to question the old wisdom and experiment in their respective fields. In fact, DeepMind has already used the algorithms that underlie the original AlphaGo to help Google slash the power consumption of its data centers.
Go, considered equivalent to chess in the West, had been resistant to machines until AlphaGo's victory. This breakthrough not only showcases the potential of AI but also underscores the exciting possibilities that lie ahead in the realm of artificial intelligence.
Read also:
- Tech Magnate Elon Musk Accuses Apple of Favoritism, as OpenAI Garners More Popular Support
- Strategies for Implementing Machine Learning in Political Campaigns
- AI Might Be the Long-Awaited Boost for African Small and Medium Enterprises - Gori Yahaya, CEO of Upskill Universe
- Uncovering the Reasons Behind Workforce Departures: Forecast Employee Turnover Utilizing Clear-Cut AI (SHAP)
