Will Google Be The First To Produce A Conscious Machine?
If we can agree that consciousness exists, then the question is whether machines can have it and if so then which company will get there first
Sitting uneasily at the junction of science, religion and philosophy - machine consciousness is the stage beyond machine intelligence, which already exists. Here are some examples of machine intelligence:
- World chess Grandmaster, Gary Kasparov, was defeated by IBM’s ‘Deep Blue’ supercomputer in 1997;
- Another IBM-developed computer, called Watson, won the TV game show, Jeopardy, in 2011. Watson answered general knowledge questions posed in natural language and bested two former Jeopardy winners;
- More recently, Apple Siri, Google Photo Search and Google’s self-driving car project are further examples of machines that exhibit behaviour that is similar to that exhibited by intelligent humans.
What is Machine Consciousness?
If we define an intelligent machine as something that exhibits useful behaviour that is understandable, predictable and controllable, then a conscious machine is not just intelligent but it can autonomously adapt and optimise within its own environment.
IBM’s Watson defeated Garry Kasparov using a ’brute force’ approach which involved running through millions of possible outcomes for each move. Watson’s ‘intelligence’ arose from its ability to apply a set of human-defined rules.
Within this narrow sphere, Watson exhibited signs of intelligence, but it was not conscious.
By way of an analogy, if a fish is fully conscious then Watson had no consciousness.
So what would Watson need to do differently for us to think it was intelligent and conscious, rather than just intelligent?
- Watson could look at the rules that humans had programmed to identify inconsistencies or gaps. It could then report back to its human operators;
- During the course of playing chess it could find an especially elegant set of moves that could become a new classic opening – Watson could make its human programmers aware of this;
- Maybe Watson discovers that the way the algorithms had been programmed was sub-optimal: perhaps working in its ‘spare time’, a conscious Watson could then make changes to the algorithms.
But most important of all, a conscious Watson would be able to take actions like these without being asked.
At this stage I should say that there is a level above machine consciousness, which is machine self awareness. I’d define a self-aware machine as one that is intelligent, conscious and can also take autonomous action to increase its own intelligence or consciousness (e.g. by analysing complex concepts such as its own existence, by showing emotion, by thinking about of other types of existence which it cannot experience directly and by wondering how its own actions are perceived).
But for now let’s revert back to the creation of conscious machines.
Is Machine Consciousness Really Possible?
Many people, including highly respected people in the tech world – like Gordon Moore, co-founder of Intel, believe that we will never be able to create a conscious machine.
I used to think that way, too, but I no longer do. Instead, I now believe that the question is not whether artificial machine consciousness will arrive, but when it will arrive and what form it will take.
I’ve changed my mind on machine consciousness for three main reasons:
- ‘Never’ is a very, very long time: Ongoing work in the fields of computer science, genetics, biology, nanotechnology semiconductors, neuroscience and neuromorphic computing is converging on the view that artificial consciousness is possible and that the first signs of a roadmap for how it could be created are emerging from the mist. Bearing in mind the tremendous technological progress that has been made in the last 50 years, I just cannot accept that machine consciousness will never be achieved;
- Consciousness need not suddenly ‘appear’ form nowhere: Before having thought carefully about this subject, I used to believe that machine consciousness would be a ‘sudden’ development: one day some company or researcher would announce that they’d created a machine that had the same level of consciousness as a human. While I believe that day will probably arrive, I also believe it will be the result of a slow, gradual progression where machines go from having no consciousness to being ‘fully conscious’. Viewing machine consciousness along a sliding scale from 0% to 100% makes it seem more attainable.
But even ‘partially conscious’ machines would transform of our world: when combined with their ability to process vast amounts of information almost in real time, even a partially conscious machine would be extremely powerful;
- The question of how life started is more fundamental than the question of whether we can create another form of life: I felt very uneasy when I started thinking that machine consciousness might be possible – because it caused me to question my own spirituality. But then I felt that machine consciousness and, beyond that, machine self-awareness just represent another phase in the evolution of the human species. I believe that we are closer to creating an intelligent, conscious and self-aware machine than we are to understanding how the first living cell emerged, or where the DNA molecule came from. However, I accept that this might just be me changing the goal posts...
Will Google be the First to Achieve Machine Consciousness?
Google is not the only institution pursuing machine consciousness, or other objectives that are on the road towards machine consciousness:
- Intel, IBM and Qualcomm are active in the field of neuromorphic computing, which uses new physical structures to replicate;
- The Geneva-based Human Brain Project has attracted €1.5 billion of EU money to deliver an electronic-software model of the human brain in 10 years;
- Although several years behind Europe’s Human Brain Project, the U.S. government has committed initial funding to the BRAIN initiative which is also working to ‘revolutionize our understanding of the human brain’;
- Along with similar activities that are ongoing in the military research departments of China and Russia, the UK and other major sovereign states, the U.S. DoD is actively working in the fields of self-piloting armed drones and in so-called killer robots;
- In addition, when viewed on a worldwide basis, thousands of researchers and engineers are working on machine learning and machine consciousness in academia, both within the public sector and in corporate R&D functions.
The consensus among the supporters on when machine consciousness will be achieved seems be between 10 and 20 years.
While that is a long time, it should be remembered that this is shorter than the age of the mobile communications industry and, certainly at 10 years, is close enough for this technology to have an effect on the valuations of companies, including Google.
The successful commercialisation of machine consciousness would represent a turning point in human development: machine consciousness would transcend all previous human inventions – combined.
While noting the other major players active in the field, at this point, we would have to say that Google is the company most likely to reach this pivotal milestone first – that is, before Apple, Facebook, Microsoft and all the other major actors in the tech space.
There are several reasons for this:
- We can already see evidence of machine learning and machine intelligence in existing Google products like Google Now, the search function that forms part of Google+ Photos and Google’s self-driving car project;
- When viewed as a technology, machine consciousness would be a perfect overlay (or underlay) that would greatly enhance Google’s core strategy, which we looked at in the previous section;
- If successfully commercialised this technology could take Google to a completely different level because it would allow the company to turbo-charge its core products and deliver major incremental value to users;
- A string of recent acquisitions in the field of robotics and machine intelligence along with the recent hiring of Ray Kurzweil as a director of engineering shows that Google is by no means done with machine learning: it is clear that the company is just getting started;
- Taking into account the present scale of Google’s computer network, the size of its user base and its capital reserves Google is well placed to take on something as ambitious as machine consciousness;
- Few, if any, comparable companies have a proven track record in investing in visionary technologies and projects that are clearly not going to have a pay-off for a long time. This again suggests that machine consciousness is a natural fit for Google.
Whoever reaches consciousness first will be able to race ahead of the competition: the victor will be able use their conscious machine to accelerate its own rate of progress. While we do not see a binary outcome where there is only one company that has conscious machines, we do see a ‘critical mass’ effect taking hold where – as with many similar ‘critical mass’ tech markets today – the further ahead the leader gets, the faster he is able to run. Eventually, competitors are left so far behind that they cannot catch up.
If there ever was a Holy Grail in tech, then it would have to be machine consciousness. Google’s development path strongly indicates that this company will be the first to find the Grail and see what’s inside – at which point everything changes.