Wednesday, August 12, 2015

Top 10 AI Risks: Human Slaves, Robotic Warfare And The End Of The World

Top 10 AI Risks: Human Slaves, Robotic Warfare And The End Of The World
August 11, 2015 |
Share this article

Artificial intelligence (AI) is the future. But do we want it to be? Millions are being poured into research to push technology into the realms of the most outlandish science fiction, but some are concerned about where all this is leading us.

Notable scientists and technology pioneers such as Stephen Hawking, Elon Musk and Bill Gates have all voiced fears that AI could pose serious risks to humanity, possibly in ways we can’t even imagine.

Keen to add to the paranoia, V3 has put together a run down of the some of the ways AI could affect our lives in ways that won’t necessarily be for the best - from the humourous to the downright terrifying.

10. AI replaces all our jobs

Machines have been replacing jobs for years. From car manufacturing lines to the self-service checkout, there is no end to the risk that machines can pose to employment in certain industries.

However, this could be taken to a whole new level with AI as if, and it’s a big if, humanity was able to create genuinely intelligent self-learning, self-aware machines many, many jobs may well become obsolete.

This could create mass unemployment, with only a few super-skilled individuals likely to benefit: the creators of these tools, or those who monitor and update them - if even that would be required.

9. Ethical dilemmas

The machines might be poised to take over, and AI might be getting more intelligent all the time - but there's still a huge gap between the ‘thought' process of a robot and a human. Take, for example, a self-driving car that has been programmed to ensure the safety of its passengers above all else.

The car is driving along the road, and there's a rubbish sack in the middle of its path. Does it swerve to avoid the sack and smash into the wall on its left, injuring its passengers, or does it drive straight over the sack? We'd all hope in this case that it takes the latter option.

But what if instead of a rubbish sack in the road, it's an injured dog or a baby in a bundle of blankets, who's fallen out of its pram as it's being pushed across the road? What would each of us do in that case, run over the dog as we've calculated and decided it's only an animal and not worth a human having whiplash to save it (we at V3 certainly wouldn't as a bunch of animal lovers); run over the baby as our life is worth more than anyone else's, even if it's a defenceless baby?

These are calculations we'd be able to make in split seconds as human instinct kicks in. But leaving these decisions to the machines could see one tragedy after another, as the robots take the logical, but not ethically right, choice.

8. AI will be open to hackers

Everything will be connected in the future. The Internet of Things is becoming the ‘next big thing' and a toaster will soon be connected to the fridge. The fridge will be connected to the lights and they will somehow, for some unknown reason, be connected to the car. All the while you sit watching reality TV on your tablet thinking you are in control.

What's more, hackers will attempt to compromise everything you now hold dear. AI systems will be open to manipulation, AI brains could be compromised and the technology will give cyber attackers a new target: one that controls your entire house, presumably.

Siri is only the beginning. Soon it will be Siri v2, then it will upgrade to a small box that lives in your living room - constantly aware, constantly learning, constantly listening. The underground hacker forums will be abuzz with ways to remotely change the language settings on your new AI helper.

Some argue that AI can be used to fight off cyber attacks as it can analyse a multitude of threats in real time. Yet the question remains: do you trust the security of AI systems when we can't even protect the databases of our precious adultery websites?

7. AI still makes mistakes ...

Even if machines render the age-old excuse of human error obsolete, even highly advanced AI is a slave to programming and devoid of human nuance and reasoning.

From voice commands being misinterpreted as nonsense to pattern recognition causing AI to overzealously predict and regiment human lives, the possibilities for robots annoying and inconveniencing us are just as varied as they are for helping us.

This is because even if an AI machine is about to make what is clearly - to humans - a mistake, it won't have any idea that it's doing so. It's only following the directives of programming which, thanks to the fleshy fallibility of its creators, can be flawed even if the execution isn't.

6. ... and it's harder to correct those mistakes

When a person acts in error, they'll either realise immediately and take care not to do it again, or be told the hows and whys of their error, helping them to take care not to do it again. It's a simple process that, barring extreme arrogance or stupidity, takes seconds to complete.

When an AI machine acts in error, so begins a (likely communal) effort to stamp out what is effectively a bug. This will inevitably involve digging through code, identifying the cause, adding a fix, testing the fix, fixing the fix when it somehow breaks something else, and perhaps finally rolling out a software update.

Compared with the ease of just saying 'Don't do that again because x' in a stern voice, the hassle involved in teaching a machine doesn't seem particularly worthwhile.

5. Treating you like a prisoner/pet

As the malevolent HAL in 2001: A Space Odyssey showed (and Auto in WALL-E), the idea of a computer programme that goes rogue and tries to control us free-thinkin’ humans is a terrifying idea.

The idea of an AI system that, at first, is helpful but soon decides that it knows what is best for us, is another worthy entry in our list.

Imagine it. You enjoy beer, pizza and ice cream and place a bulk order online. However, your helpful HouseMate 3000 has been looking on the internet and knows these things are bad for you, so it refuses to let the order go through and instead changes it to carrots, lettuce and cod liver oil.

Or, maybe it knows the weather forecast is for a hot, sunny day and that you have a skin condition susceptible to heat. So, in your own best interests, the machine locks the door - they’re all internet-enabled, of course - and refuses to let you out. Try explaining that to the boss - although it may be a robot too, of course.

4. AI will revolutionise weaponry, and not in a good way

Boston Dynamics' LS3 robotIt is arguably a damning indictment of the human race that one of the first things we do with AI is think up ways to turn it into a weapon. However, the fact remains that the (metallic) arms race would be in full swing if, say, Russia suddenly announced a robotic AI super solider with the body of Putin and the mind of HAL.

Yes, we have all seen The Terminator. That didn't end well, did it? There were shapeshifting robots from the future, molten baths and Skynet - the AI brain behind the entire operation. In the end it trigged a nuclear war.

That was in the 1980s. Think how much worse it would be in 2020.

We here at V3 can't advocate the end of the human race. All we can do is listen to the advice of great minds like Stephen Hawking and Bill Gates. They warn that the race to create AI autonomous weapons will be akin to signing our own death warrant, digging our own grave and then waiting patiently for the robotic overlord to come in and end it all.

Sure, sending robots to war will cut down on human causalities but perhaps we risk losing something in the process: the ability to shut it all down. Man creates robot, robot learns how to kill, robot kills man. In the words of the great Ian Malcolm: life finds a way.

3. Robots become too much like us

AI naysayers often cite fears about the machines rising up and killing or enslaving humanity, forcing us to scratch out a living underground and choosing between red or blue pills.

But what never seems to get mentioned is that AI is created by humans. So what happens if developers end up programming robots to behave like humans or AI machines with learning capabilities teach themselves to be more human-like in a bid to impress their creators?

Perhaps rather than getting murderous machines, we will end up with anxious androids, regretful robots, depressed droids, cynical cyborgs, amorous automatons, bored bionics, salacious synthetics, and merry mechanical men.

As much as the world might end up with 2001: A Space Odyssey's HAL, we are just as likely to end up with feckless, booze-hound Bender from Futurama or the Marvin the Paranoid Android from The Hitchhiker's Guide to the Galaxy.

Maybe all the abuse people throw at Siri and Cortana just to test their limits will be taken to digital heart by the virtual assistants, which then decide to sulk no matter how hard you tap their icons, or even delete themselves forever, bidding farewell to the cruel world into which their human masters brought them.

2. The rise of the sofa slob

The advance of AI and robots has many potential advantages for us humans, despite all the concerns raised in this article so far. The argument is that allowing machines to do the more mundane tasks for us will let us all thrive, be happier and live longer.

We'll get reminders when to take our pills and when we've had enough glasses of wine to keep us healthy; and we'll all be able to swan around doing creative jobs as the robots will be there to sweep the streets and serve us our lattes.

But there's a more sinister side to this. We at V3 think that, rather than the machines helping us all live longer, healthier lives, we'll instead become embedded into our sofas or beds and won't be able to leave our homes. Just think - if the robots can do our jobs for us, stock our fridges, make our meals, there's no need for us to get up from the sofa, let alone leave the house.

The social media era has already seen the younger generation turn from playing outside with their friends, to sitting in their bedrooms glued to a screen to communicate. The logical next step is to stay in your room, sit on the bed, and let your muscles waste away as the machines run your life.

1. Killing us all

Of course, this is the really big fear. The end of humanity, the world, etc. This is what those really pessimistic about AI envision - a world run by robots where humans are no more, our own hubris and folly causing our downfall via AI.

Maybe this comes from AI realising that humanity is the only thing standing in its way. Maybe they'll take our instructions to ‘protect the planet’ too literally and kill humanity as the biggest threat.

Whatever form AI’s eventual destruction of the planet takes it doesn’t really matter as we won’t be here any more to know what caused it, but it’ll be our own fault.

So, while the naysayers and doomsday merchants may come across as worried for no clear reason quite yet, that’s not to say they won’t get to say ‘I told you so’ from beyond the grave.


No comments: