Listen to the full interview with A.I Expert Jakob Foerster below, or read on for a summary.
In 2014, Google purchased the British Artificial Intelligence start-up, DeepMind, for a massive £400 Million. Shortly after the acquisition Deepmind's founder, Demis Hassabis, took some time away from being a newly crowned super-millionaire to unveil his two step plan for the company -
1) Solve Intelligence
2) Use this knowledge to solve everything else
Pretty vague stuff there, Demis... then again, if I had a new vault of coins to swim in, I wouldn't want to be wasting time at a press conference either. Nevertheless, this is the very plan many Machine Learning, and A.I. experts around the world are now working to. What is intelligence? And how do we replicate it.
Whilst working for Google, Jakob Foerster found himself sat in a room listening to Demis give a speech. He realised that his previous area of study, Physics, seemed to have a lot of the ‘big questions’ already answered for him. A.I and Machine Learning had so many new and exciting discoveries to be made. Discoveries that would propel mankind into a much simpler future. (We hope.)
A number of years (and a swift career change) later, Jakob was working on the recent adaptation of Google DeepMind’s A.I called ‘deep distributed recurrent Q-networks,’ (DDRQN). His work was well documented in the press – ‘A.I had Solved Google’s 100 Hat Riddle’. Although this notoriously baffling riddle long predates Google, it was associated with the company due to it being used to identify the highest calibre applicants during interviews. So it's really, really hard, OK?
To solve the riddle the DDRQNs began to communicate information back and forth, and as a result, worked together to come to the correct solution. But here's the clincher - they did this without having a 'pre-determined communication protocol'. In layman’s terms, the computers spoke, without being told how to speak. Not in a discernible language, but by relaying small amounts of digital information between themselves.
Computers that communicate on their own accord! What's next?! World domination, right? ...Well, that’s what the media would have us believe.
It’s not the dawning of a fire-scorched earth and Schwarzenegger replicas, quite yet.
It’s almost impossible to find a positive article, video, or information of any kind about A.I. The internet is filled with scaremongery, the newspapers with lavish headlines proclaiming a ‘bleak future’ for the human race with the entrance of A.I, and great minds like Stephen Hawkin and Elon Musk acting as prophets of doom for the technology. But the people who actually work in the field – like Jakob – are confident in the benefits of A.I, proud of their work, and realistic about the rate of progress.
Sentient computer programs are "a long way off’ says Jakob. "The computer didn’t sit down, read a riddle, think about it over a cup of coffee and produce the answer", there’s still an essential human element for the computer to do what it does best; a computation. Of course, saying this should not undermine the incredible, and groundbreaking work his team have done…but the reality of the situation wouldn’t make great headlines. Don't panic, we aren't on the doorstep of a fire-scorched earth and Schwarzenegger replicas, just yet.
Yes, we may be a way off a sentient A.I, but the technological developments that have already taken place, and those that will exponentially increase over the next decade, will probably change our lives as we know it. Albeit, slowly. Probably slow enough for us not to really notice. It will just become our normality. “We won’t wake up and all of a sudden there’ll be a sentient being. When we get close, we’ll know.”
In our conversation with Jakob we cover everything from the already life-consuming ‘Intelligence Augmentation’ to real world A.I developments, the prospect of ‘Peak Technology’ and whether it is possible to artificially replicate creativity in it’s truest form.
The developments in the field are more impressive by the day, and understandably hard to stomach for some of the population. But when you hear it all come straight from the horse’s mouth rather than through a sensationalist, apocalypse-laden media filter, A.I seems more intriguing, and dare I say, useful, than it does daunting.