AI&SOCIETY - 21/02/2020

Google’s Acquisition of DeepMind, Five Years Later

Google entirely surprised the public when it announced the purchase of British startup DeepMind back in 2014. For a staggering £400million, it was the most expensive acquisition ever made in Europe. 

DeepMind, a small but ambitious company, was initially founded in 2010, with the focus on advanced AI technology. Since 2015, it has been officially owned by Google’s parent company Alphabet and has research centers in the US, Canada, and France.

It has continued its growth in the field of cognitive systems, intending to develop superior Artificial General Intelligence that will be broader than the human mind and ready to handle even the most complicated tasks.

Where Is DeepMind Now?

So far, we saw some quite fascinating ventures performed by the company. But such a considerable investment requires years of a successful business to recover the invested money. Yet, after five years, we can note some significant breakthroughs that the company has made.

AlphaGo

In October 2015, DeepMind created artificial intelligence capable of competing against humans in playing games. The computer program AlphaGo played the first game against European Go Champion, Fan Hui. Never before has a machine been able to handle so many possibilities like with game Go. Still, AlphaGo won with the result five to zero. 

Over the next two years, the machine managed to defeat world champions and record holders such as Lee Sedol and Ke Jie. In the end, with the analysis of the massive number of parties played among humans, AI, based on the deep reinforcement learning, become almost invincible.

AlphaFold

Between 2016 and 2018, the company worked intensively on AI protein folding, one of the most challenging science areas today. DeepMind developed a project that successfully predicted 25 of a total of 43 proteins, which represented a revolutionary and pioneering breakthrough for AI technology.

WeveNet

In 2016, the company introduced a text-to-speech cognitive system. Google soon implemented it into one of its services, Google Assistant, and later developed a similar commercial version, Cloud Text-to-Speech.

AlphaStar

AlphaStar is a computer program that uses a reinforcement learning system to learn from the previously played StarCraft games, but also by playing against itself. After the training program, the machine had 200 years of game experience and was more than ready to play against human opponents. By October, it reached the Grandmaster level.

DeepMind In The Future

It looks like Google made an excellent merger at the right time. Although the numbers say otherwise, AI technology could keep this tech giant in leading position for many more years. With the announcement of revolutionary healthcare programs, DeepMind could develop technology to diagnose, even treat many diseases, but also provide better patient-healthcare interaction and save millions for the industry.