Google’s DeepMind develops advanced AI program
Google’s artificial intelligence division, DeepMind, has created an advanced AI program that can teach itself beyond the human information it’s initially given. Google shares ended Thursday trading little changed and looks likely to open the US trading day in the green.
The Al program is called AlphaGo Zero and is programmed to play the ancient and still popular Chinese board game, Go.
The program was only given the rules of the game and in just three days mastered it and created moves of its own.
Significant development
While the ability to master a board game in three days is impressive, you’d be forgiven for thinking it’s not particularly important. But you’d be wrong.
The team that works on AlphaGo Zero, are based in the DeepMind London facility and they’re aiming to create a more general-purpose AI system. One that can be used to learn and operate in a number of different ways. Right now, much of the AI that’s available is quite narrow and only able in one area.
"We're quite excited because we think this is now good enough to make some real progress on some real problems even though we're obviously a long way from full AI," DeepMind’s Chief executive Demis Hassabis told journalists.
Google at centre of US election investigation
While the AI breakthrough is being hailed as great news for the tech giant, its investors and consumers, Google shares are still to recover from the hit they took Wednesday, following news the business was involved in ad targeting during the US election.
Both Google and Facebook worked with an advertising agency representing Secure America Now. In the run up to the 2016 election, they assisted them in better targeting their ads and message to the right audience.
After working on the project to target ads more strictly to a certain group, Google said it eventually blocked some of the ‘faux-tourism’ ads. The ads portrayed a strong anti-refugee message that were intended to be shown to audiences who would be reasonably influenced by them.
