Google’s DeepMind poised to start making money from AI

DeepMind, Google's London-based artificial intelligence (AI) company, finally looks to be realising the potential for commercialising its machine learning research and development through embedding AI into a host of Google products and services.

The research Google's DeepMind has done over the past year has helped the company improve its The Google Assistant, the search giant's voice-controlled virtual assistant, making it sound more natural as demonstrated at the Google event earlier this week in products such as the Pixel 2 phone and the new Google Home Mini.

DeepMind, which Google purchased in 2014 for $400 million, focuses heavily on AI research including teaching AI to beat humans at board games. While this is incredible research, DeepMind's ability to make a profit off of it has proved to be difficult. In fact, in its first year, the company reported zero revenue.

DeepMind announces second NHS partnership Google Pixel 2 hands-on review: Back and better than ever What is AI?

However, last year, DeepMind developed an algorithm, WaveNet, which according to the company's website, is "a new deep neural network for generating raw audio waveforms that is capable of producing better and more realistic-sounding speech than existing technique".

While originally thought to be "too computationally intensive", DeepMind improved this algorithm within the last year making it 1,000 time faster and allows it to produce higher fidelity sound, allowing Google to utilise it in its new products, as reported by Bloomberg.

While the company is still spending more than it is taking in, reporting a loss equating to 94 million just last year, vin part due to hiring exprienced and technical staff to fuel its AI development and research.

However, its increasing involvement in Google's headline products and services could eventually lead it the company making a profit.