Google’s DeepMind becomes ‘highly aggressive’ when stressed warns search giant
Despite numerous predictions from experts that AI could lead to the downfall of humanity, including Stephen Hawking who claimed it could be the “worst thing” for human beings, tech giants are continuing to invest heavily in AI.
In 2016, Google’s AI program, known as DeepMind, showed its makers it was capable of learning independently, teaching itself to beat the world champion in a game of Go!.
Now, it has continued on its ruthless streak and opts for “highly aggressive” strategies when it is in fear of losing.
In the latest tests, two DeepMind agents were tasked with playing a game of ‘Gathering’ – a computer game where two people, or in this case computers, play against each other to collect the most apples.
DeepMind's AlphaGo beat the world champion of Go!
The AI beings operated smoothly when there were enough apples to go around, but once the apples became more sparse, the DeepMind systems began using laser beams, or tagging, to knock the other one out, ensuring they could collect all the apples.
A blog post from the DeepMind team read: “We let the agents play this game many thousands of times and let them learn how to behave rationally using deep multi-agent reinforcement learning.
Will 2017 be the year of Artificial Intelligence?
Thu, January 5, 2017
Robotic assistants are on the rise, here's how AI is getting more human.
1 of 9
Get Quotes on Home Insurance
Asus Zenbo: This adorable little bot can move around and assist you at home, express emotions, and learn and adapt to your preferences with proactive artificial intelligence.
DeepMind mimicked behaviour of human-learning
“Rather naturally, when there are enough apples in the environment, the agents learn to peacefully coexist and collect as many apples as they can.
“However, as the number of apples is reduced, the agents learn that it may be better for them to tag the other agent to give themselves time on their own to collect the scarce apples.”
There are fears that AI's wishes will not fit in line with humanity's
Joel Z Leibo, who is part of the DeepMind team, told Wired: "This model… shows that some aspects of human-like behaviour emerge as a product of the environment and learning.
"Less aggressive policies emerge from learning in relatively abundant environments with less possibility for costly action.
“The greed motivation reflects the temptation to take out a rival and collect all the apples oneself.”