Senior lecturer, Faculty of Science, Engineering and Built Environment, Deakin University
Read more
#1 Victorian uni for graduate employment1
#1 in the world for sport science2
#1 Victorian uni for course satisfaction3
Twenty years ago the ultimate man vs machine matchup played out. On one side was humanity’s best: Garry Kasparov, arguably the greatest chess player of all time, and the as then world champion. Opposing him was Deep Blue, a chess playing supercomputer built by the engineers at IBM.
By beating Kasparov in the most human pursuit of playing a game of intelligence, chess, Deep Blue ushered AI onto the world stage and into the public’s consciousness. Since that day in 1997 debate has raged over the match’s significance. Tragedy? Achievement? Even now, opinion is still divided.
On the one hand we weren’t the most intelligent beings in the room. On the other, it appeared that ‘intelligence’ wasn’t all it was cracked up to be. Was this the beginning of the end for humanity as the computers took over?
Anxiety about AI and robots appears to be a natural viewpoint for humans, from the first appearance of the term ‘robot’ in the 1920’s Czech play, Rossum’s Universal Robots, where a race of robotic servants wiped out humanity, to James’ Cameron’s synthetic intelligence ‘Skynet’ in the Terminator franchise. Was the match between Kasparov and Deep Blue the moment where our anxiety was justified?
When we think of chess, we think of intelligence. When professionals play, they plan, analyse, evaluate, and learn. So, testing a supercomputer’s ability in such tasks made chess a natural choice.
However, according to Dr Tim Wilkin, Senior Lecturer in the School of Information Technology at Deakin University, there was a far more practical reason that chess was chosen. ‘Playing chess is a matter of searching very efficiently through the many possible sequences of moves available given any arrangement of pieces on the board, to find a sequence that means I win and you lose,’ Dr Wilkin says.
‘Deep Blue was more about testing our ability to build machines that were highly efficient at solving search problems, than it was about proving machines were more intelligent than humans. If you’d asked Deep Blue to make Kasparov a cup of tea after the match, it couldn’t even process the request, let alone fulfil it.’
However, as Dr Wilkin explains this is no simple task in chess. ‘The average number of branches at any point in the tree is about 35. This means that after just five moves by each player, the number of states in this sub-tree is more than 35(10). That’s about 3 quadrillion (3 billion billion) possible game states to evaluate.’
If we consider that the average chess games are about 40 turns long, or 80 layers in the tree, then we find that a computer cannot consider all move sequences to find an optimal solution. It must use ‘intelligence’ to consider only good moves that are likely to lead to a win.’
Dr Wilkin explains that the approach taken by Deep Blue was to apply an ‘evaluation function’ to the board state, which rates the quality of that state for the player about to make a move.
‘In building its search tree, Deep Blue uses the evaluation function to favour searching move sequences that minimize the opponent’s chance of winning. In a zero-sum game, that maximises the computers chance of winning’.
The true intelligence in Deep Blue is contained in the evaluation function, which was designed by analysing millions of recorded chess games, as well as ‘hand crafting’ by humans and other chess champions. Ultimately then, Kasparov wasn’t simply playing a machine, but a whole team of human chess experts.
Some of the most intelligent people in the world have warned us about the development of AI and what it means for humanity. Recently, Stephen Hawking, Bill Gates, and Elon Musk have warned that the development and creation of AI has the potential to bring about the destruction of the human race through the creation of super intelligent machines that will see no use for us.
Dr Wilkin argues that while this is one possible future, it’s far more likely that it won’t be the case. ‘When Deep Blue beat Kasparov, it wasn’t machine intelligence that enabled it to win. It was the marrying of Deep Blue’s computational power, with the shared experience and knowledge – what we would term intelligence – of IBM’s engineers and the other world championship level chess players they used to build its evaluation function.’
'Playing chess is a matter of searching very efficiently through the many possible sequences of moves available given any arrangement of pieces on the board, to find a sequence that means I win and you lose.'Dr Tim Wilkin,
Deakin University
‘We look at what Deep Blue meant for the world and we basically see the growth of the internet from that. We had the technology to build and network computers, and the ability to store knowledge, but Deep Blue’s architecture and that of later systems has allowed companies such as Google to create search engines capable of analysing vast quantities of stored data in short time periods. The outcome of this is that society no longer values the acquisition of knowledge; we now value the ability to use knowledge. That’s an interesting twist,’ Dr Wilkin continues.
With cyborgs becoming more commonplace, this could be our next evolutionary step. The combination of our intelligence, with the computational power of AI means that we could follow more human pursuits, while leaving the dangerous tasks to robots and AI.
‘Now we have jobs that require creativity, innovation and design thinking, as opposed to thirty or forty years ago where many jobs were focused on manual labour, or retained knowledge. Now we are looking for people to use knowledge, and use it more creatively and entrepreneurially,’ Dr Wilkin explains.
Dr Wilkin argues that we only have to look at the development of past technology, and the benefits it brings to humanity, to potentially see the benefits of AI.
‘The evidence is that we’re much better off for many of our technological developments. We have longer life spans, we’re able to stave off diseases and repair bodies, we produce food and energy more efficiently and the standard of living is higher for many. By creating intelligent machines and intelligent software, which will take care of many laborious physical and cognitive tasks, it’s likely that we’ll be able to pursue more intellectual and creative activities,’ Dr Wilkin says.
So, while it’s always good to be cautious about what we create, it’s not all doom and gloom. We should also look at the massive benefits that the development of AI can bring to the human race. Or, as Dr Wilkin says; ‘Just like Deep Blue wasn’t the start or the end for chess, I don’t think AI is the end. I think it’s another possibility for us.’
Want to understand the role of AI and how humans will adapt in the future? Study a Bachelor of Computer Science at Deakin University
Senior lecturer, Faculty of Science, Engineering and Built Environment, Deakin University
Read more