Artificial Irrelevance: The robots are coming

18 Jul 2012, 6:15 pm   -   Source: Andy Park, SBS
Share This
9 
+ Comment
8

The creation of artificial intelligence more sophisticated than a human brain, even by accident, is now a very real prospect warns Skype co-founder Jaan Tallinn.

RELATED

Science fiction luminary Isaac Asimov’s Three Laws of Robotics mandated safety measures to limit artificial intelligence in robots, protecting humans from harm.

That was 70 years ago. Largely written for fantasy books.

Nowadays, researchers have developed artificial synapses, computers have won jeopardy and Google's computer can do something called supervised learning.

With the robotics industry saying household robots will be the new PCs in ten years, the impact of robotic AI on humanity is no longer tomorrow's problem.

RISKS OF SINGULARITY

Experts say the principles that govern AI are dangerously insufficient and even expose humanity to "extinction-level risks".

In short, the possibility of humanity creating a technology system more sophisticated than a human brain, even by accident, is now a very real prospect.

Futurist Ray Kurzweil, more recently obsessed with immortality and nano-medicine, popularised the term "singularity" for that moment.

Some are calling for a narrowing the capabilities of AI, before private industry or militaries develop an AGI or Artificial General Intelligence, that is greater than ours.

AI vs EVOLUTION

Founding engineer of Skype and co-creator of the pioneering file sharing service Kazaa, Estonian programmer and philosopher of technology Jaan Tallinn is at Sydney University to talk about his ”Intelligence Stairway” theory.

He argues that artificial technologies will, and is some cases have already, taken over from biological evolution, rapidly propelling us to an intelligence explosion.

"Evolution is an optimisation process." Jaan says.

"It is trying to optimise the fitness of organisms against a background of environment and (that's) also like machine-learning," he said.

"Evolution actually made a sort of mistake in a sense that it actually created primates with optimisation ability and that optimisation ability got powerful enough to actually understand evolution. It actually created something that was more powerful than itself."

Jaan's point is that humans are on the verge of potentially repeating that mistake.

"I’m giving about 50 per cent probability of this thing (technological singularity) happening this century," he said.

"If this thing is going to be really slow, then people have time to turn this thing into policy and then because this is a really contentious issue, it might end up doing a lot of damage," he said.

He also warns of the dangers of only seeing AI as Hollywood's humanoid robots.

"I mean intelligence is really about planning and protection and you don’t really need arms and legs to do that. For example, Google is a very famous application of AI."

CYBORGS: A SLOW INTELLIGENCE EXPLOSION

Oxford University's Dr Anders Sandberg is more circumspect about singularity and the rise of the machines - but for somewhat unnerving reasons.

"I don’t think it is very likely, that they get out of hand and subjugate us and wipe us out," Dr Sandberg said.

"We’re made out of very useful atoms that can be configured to something else,"

For someone with a background in computational neuroscience, Dr Sandberg now spends more time on the philosophy behind technology-enabled collective intelligence, a sign of the growing coalescence between the once disparate fields of computing and philosophy.

Dr Sandberg is one of a unique group of thinkers at Oxford's Future of Humanity Institute who have been set loose on the big-picture questions, largely the future of intelligence.

"What intelligence is and how to achieve it has turned out to be a surprisingly hard problem to solve," Dr Sandberg said

"As soon as we achieve [it] then we see people say "oh that’s not real intelligence" he said.

Dr Sandberg mediates a more harmonious "intelligence explosion", where a slow blend of intelligences might occur.

"We would have a situation where a total amount of intelligence rising exponentially but still on a time scale that allows us to control it or various forms of intelligence control each other," Dr Sandberg said.

He predicts that a non-human intelligence, created in our likeness, would share our values.

"Think about how our society works," Dr Sandberg said.

"We have a lot of minds here and not all of them are nice and some of them want to subjugate other minds but we control that by norms, by ethics, and by good upbringing and having police and institutions and economic incentives to actually behave nicely,"

"If you have a growth of smart machines in the same way, well then you can just integrate them so they will also not want to run afoul a police and thrown out of the community because it’s so beneficial to be part of it," he said.

Some AI experts, like Marvin Minsky, have begun to approach emotions as logical ways to think in addressing different types of problems.

A CAUTIOUS APPROACH

Even Jaan Tallinn admits that we know so little and what do so know changes almost daily.

"There are all sorts of possible scenarios out there. Some of them are just disastrous; some of them are like Utopian, really good. And we have to be aware of the whole spectrum of things and not get fixated on the utopian side of things and think about everything that can happen," Jaan said.

Leading AI programmers now talk of a superhuman partnership or complementary computing, rather than a super-intelligent competitor.

Foundational computing engineer John Irving Good wrote in 1963 that "the first ultra-intelligent machine is the last invention that man need ever make."

But if super-intelligence is created with all the flaws and emotions that bind us, that may only be the case if our craftsmanship is not perfect.

Robots , Andy Park SBS

  • Follow us: 
  • Follow us on Facebook
  • Follow us on Twitter
  • Follow us on Youtube

Your Comments

Are we already there?

Locked in - from Alice Springs, NT, 7 months ago

As I find myself more and more reliant on smart phones, google and wi fi, I wonder are we already being superceded by technology? It is important to consider "AI" as not just a humanoid robot, or a large computer in a warehouse. AI is all around us right now, in the form of an incredible evolving complex network - the internet. It is conceivable, sometime in the future, for the internet to work at speeds close to human brain synapses, therefore creating the ultimate brain / conciousness

Flick a switch, turn it off!

Aree Cohen - from Sydney, 10 months ago

The scientist proposing this doomsday scenario, are clearly "materialist" in their basic understanding of the world we live in. Intelligence, is a tool of consciousness (I think therefore I am). There's absolutely NO science that supports the idea that consciousness is the outcome of highly evolved material processes. Setting up an academic task force for this, is typical of academics who are increasingly pressured to churn out publications regardless of merit. Have we forgotten the Turing Test?

Accidents will happen but only in Hollywood

Niccolo Capanni - from Aberdeen, 10 months ago

"the possibility of humanity creating a technology system more sophisticated than a human brain, even by accident,...." More sophisticated is possible but not by some movie-like mad scientist accident. People are working very hard to achieve this on purpose. Sophistication does not equate to Artificial Intelligence. While we have AI immersed in current technology, it is a big step to hard AI, singularity, or general AI (much the same thing) and we won't get it just by increase sophistication.

Our children will have to deal with this issue...

Chris - from Weston Creek, Canberra, 10 months ago

...but we've got many years before robots are walking, working and living among us. Hopefully, it will work out well. http://technyou.edu.au/2012/07/thinking-robots/

Interesting

Frank - from Melb, 10 months ago

Interesting thought, if this was going to happen... AI becomes smarter than humans, what will motivate AI, if Humans are running off instinct (mainly driven by sex) what will drive AI? Will it go to any level to survive, or will it relies it's pointless existence and simply shut down...

None

Herman Cummings - from Fortson, 10 months ago

If pastors, priests. rabbis, and "so called" Christians would stop their false (old Earth) and foolish (young Earth) teachings, and start promoting the truth of Genesis (Observations of Moses), then there would hardly be any room for the ridiculous teaching of evolution. Collectively, Bible believers are so "blind", that their approach to Genesis is a joke. Instead of seeking the truth, they continue to support the current lies and foolishness of Creationism.

Great quote on evolution

Norboden - from Minneapolis, 10 months ago

Artarmon, I would suggest that we are actually better at optimization then evolution not worse. Think of the speed with which we are now able to adapt biology. In 100 years we can eliminate a disease while normal evolution might take 100,000 or even millions of years. If an AI had a similar speed increase it would start changing life as we know it from one day to the next. Unimaginably paced change may face us in the future with the pace dictated by our ability to utilize it effectively.

Great quote on evolution

Sydney Artarmon - from Artarmon, Sydney, 10 months ago

"Evolution actually made a sort of mistake in a sense that it actually created primates with optimisation ability and that optimisation ability got powerful enough to actually understand evolution. It actually created something that was more powerful than itself." Interesting way to put it. However, evolution is still better at optimization then we are, as we now use evolutionary algorithms for optimization. Could an advanced AI mean a further reduced version of this gift?

Join the Discussion

Name
City / Suburb E.g. Artarmon, Sydney
Title
Comment
You have characters remaining.
Validation
What's this?
This is a captcha-picture. It is used to prevent mass-access by robots.
All submitted comments become the property of SBS. They are moderated, so we reserve the right to edit comments and remove HTML tags. Not all submitted comments will be published. Publication does not mean we endorse the opinions expressed. Please read our terms and conditions for more information.