instagram envelope_alt facebook twitter search youtube_play whatsapp remove external_link loop2 arrow-down2

Connect: Room with AI view

AI is all around us — but is that a good thing? From Google Duplex to ‘co-bots’ METRO investigates whether IT is humanity’s friend or foe…

AI HAS a PR problem. In almost every sci-fi film of the past 50 years, artificially intelligent beings have attempted to wreak havoc with humanity. And while there are notable exceptions — we’re with you, Wall-E — the overwhelming feeling from Hollywood is that a future with AI won’t be a happy place.

But AI isn’t the future — it is the now. Without us even realising, artificial intelligence has embedded itself into our lives. From what we watch on TV to the pictures we snap on our smartphones, AI in its various guises is already an essential part of our every day. So is AI something to be afraid of or to embrace? And do we really have a choice?

AI 101

Artificial intelligence is really an umbrella term for a number of different technologies and concepts — such as machine learning, deep learning and natural language processing — that describe how computers try to think and learn like humans.

Predictive texting is a simple example of an AI often called narrow AI. At the other end of the spectrum is so-called artificial general intelligence, a holy grail for many AI researchers. Here the aim is for machines to exhibit intellect, intelligence and an ability to ‘think’ like humans.

AI warning: Elon Musk

This area of research isn’t without controversy, with high-profile figures including Elon Musk and Stephen Hawking having warned that the development of full AI ‘could spell the end of the human race’, a process referred to as ‘technology singularity’.

However, experts say that AI is less about machines taking over from humans and more about intelligent systems standing shoulder to shoulder with us, helping humanity to be the best it can be.

Imitation game

Chatty AI assistants like Siri, Alexa and Google Assistant have become ubiquitous in recent years, allowing us to command our refrigerators, microwaves and even toilets with our voices.

They are becoming increasingly common outside of our homes too. McDonald’s customers ordering at drive-throughs and in-store kiosks may soon find themselves asking a voice-enabled AI for a Big Mac and fries after the company recently bought customer-service AI start-up Apprente. The fast-food chain even hinted that the AI assistant could appear in its own smartphone app.

Chatty: Assistants such as Alexa are increasingly common

While these AI assistants may have plenty of digital tricks up their sleeves, they are still a long way from fooling anybody that they are human. However, in 2014 a conversational AI called Eugene did exactly that by becoming the first computer program ever to pass the Turing Test.

Conceived by code-breaking British computer scientist Alan Turing in 1950, this is the test in which humans attempt to distinguish between a computer and a person during a text-only conversation. The victorious AI, which simulated the vocabulary and conversation of a 13-year-old Ukrainian boy, convinced a third of the human judges that they had in fact been conversing with a real person.

But the AI that has attracted perhaps most attention — and controversy — recently is Duplex. Google’s voice assistant is able to make telephone calls on your behalf to book restaurants, hair appointments and even hire a car. Its human-like voice and conversation skills sound so natural — including ‘um’ and ‘er’ hesitations — that in trials people on the other end of the phone had no clue they were in conversation with an AI. Duplex is already available to smartphone users in the US and the web version will roll out to UK users later this year.

Exterminate?

Hal 9000 [from 2001: A Space Odyssey], Terminator and Ex Machina’s Ava are A-listers among a long Hollywood cast list of malevolent machines that portray AI as an enemy of humanity and certainly not its willing assistant.

However, far from attempting to exterminate us, artificial intelligence is now being used to help create new human life. The AI behind the Apricity Fertility Predictor has been trained on half a million cycles of anonymised fertility data from the UK’s Human Fertilisation and Embryology Authority and, given the prospective parent’s age and medical history, can predict the estimated chances of conceiving using different fertility treatments.

Hollywood sci-fi: Ex Machina was among films to portray AI as the enemy

AI is helping to prolong human life too. IBM’s AI engine Watson has absorbed every medical research paper, journal and case study ever published — far more than a clinician would ever be able to read, let alone recall — to support healthcare professionals’ diagnoses.

Google has been training its deep-learning systems on early detection of lung cancer and skin disease, claiming detection rates similar to, and even better than, human experts.

And while a common fear is that AIs will render humans redundant in many jobs, ‘co-bots’ — smart, collaborative robots that work alongside people — are now being used by car manufacturers such as Nissan and Ford, performing heavy lifting and repetitive tasks.

Screen stars: Hal 9000

TV AI

Just as our smartphone screens have expanded to fill our pockets, so too have TV displays grown to fill our living rooms with lavish canvases fit for glorious high-definition content. But the sizeable screens come with sizeable challenges many manufacturers are addressing using AI.

To prevent blocky blockbusters and allow box sets to make the most of the screen estate available, TVs often employ some form of upscaling that adds in the missing pixels to fill up the screen.

That’s where AI comes in. For example, some of Samsung’s newest TVs employ a Quantum AI upscaling processor, which uses machine learning to identify what is about to appear on the screen and compare it with an onboard memory bank of similar images and scenes.

It then uses settings and filters from this bank to intelligently upscale and fine-tune the image in real time, resulting in a picture that looks like it was shot for the big screen.

Computational photography

Where the laws of light and physics may limit the capabilities of our phones’ minuscule cameras and lenses, AI is helping to create pictures that would otherwise be possible only with sensors and glass many times larger and heavier.

Google has stuffed some of its AI know-how in its Pixel smartphones to great effect, adding features such as Super Res Zoom. But so-called computational photography goes much further than just mimicking big camera kit.

New colleagues: Co-bots work alongside people on car-making production lines

What was once a minefield of camera settings is no longer a concern. AI-powered scene selection and post-processing seamlessly stacks multiple exposures, adjusts lighting and skin tones, and even adds bokeh-style blur to backgrounds without breaking sweat — steps that once would have required specialist software and a ton of time to accomplish.

So is AI something to fear? It’s a ferociously complex beast. Still, it’s not so much how we use it but how we teach it that counts.