Deep in thought

For those who have taken an interest in National Science Week, you may have noticed that I have not responded to this year’s theme – ‘Deep Blue’. This is partly because the reference is to ‘the future of oceans’ – and I have not been particularly inspired by artworks in the National Gallery of Victoria collection. But it is also because I have been struggling with the word ‘deep’ which seems to be another ‘mot du jour’ – another word which has gained ‘unprecedented’ popularity.

We hear that when environments have been exposed to the coronavirus, they need to have a ‘deep clean’; that when topics need to be seriously considered we take a ‘deep dive’;  that in the economic downturn we can’t expect the government to have ‘deep pockets’; that the inability of people to conform to safe practices will land us in ‘deep water’; that we all need to ‘dig deep’ to support each other; and that America’s social and political issues might be due to the ‘deep state’ conspiracy!

In 1989 ‘Deep Blue’ had an incarnation as an IBM chess-playing computer that became famous for winning a game against world champion, Garry Kasparov, seven years later. In the current context of linguistic appropriation, I wonder whether ‘deep blue’ might come to mean the distress and mental health issues people face during lockdown – a step on the way to Beyond Blue.

The word ‘deep’ has a number of meanings but in the current context it has come to connote both ‘very strongly felt and usually lasting a long time’ and/or ‘needing serious thought or difficult to understand’. Given the ‘unprecedented’ circumstances it is not surprising that ‘deep’ has gained widespread currency.

Even the National Gallery of Victoria has embraced the phenomenon of ‘deep’ with its appropriately deep blue recent online publication ‘Deep Read’ – A collection of narrative non-fiction from its Magazine which is now available online: https://www.ngv.vic.gov.au/ngv-magazine-online/deep-read/

However, the reason I am focussed on ‘deep’ during National Science Week is because of the recent rise of a phenomenon known as ‘deepfake’. Deepfake (a portmanteau of ‘deep learning’ and fake) is part of a world of synthetic media in which the image of a person in an existing photograph or video is manipulated or replaced with someone else’s likeness. It is best understood as ‘photoshopping on steroids’.

Using a form of artificial intelligence (AI) called deep learning, deepfaking uses either AI algorithms called encoders and decoders, or a generative adversarial network with generators and discriminators, to create utterly realistic but completely faked images.  An example that is often used to illustrate this is the short video of Barak Obama describing Donald Trump: https://www.youtube.com/watch?v=cQ54GDm1eL0&feature=emb_logo

Deepfakes emerged on the website ‘Reddit’ in 2017 and have since proliferated at an extraordinary rate. An article by Ian Sample (Science Editor for The Guardian) in January this year pointed out that the AI firm Deeptrace had found 15,000 deepfake videos online in September 2019 – a near doubling over nine months. While the majority of the images were pornographic, deepfaking has also been used in cyber threats, and in the political world to destabilize political processes through cover-ups and smearing.

At the current time, making a good deepfake requires considerable expertise and high-end computers with powerful graphics cards – although on the website deepfakesweb.com you can have a go at creating your own short personal video for around $10.

As the technology evolves and becomes more accessible, it is anticipated that deepfaking might create legal problems if, for example, faked events are used in child custody battles or employment tribunals. Deepfakes may also pose security risks and create scam possibilities if they mimic biometric data.

Fortunately, it is not all bad news. Voice-cloning deepfakes can restore voices which have been lost due to disease. And deepfake videos are being used in galleries and museums to enrich the experience.  The Dalí Museum in Florida uses a deepfake of the artist to introduce his art and provide an opportunity for a selfie with Dalí: https://www.youtube.com/watch?time_continue=130&v=BIDaxl4xqJ4&feature=emb_logo

 We are used to science being based in empiricism which emphasizes the role of experience and evidence, especially sensory perception, in the formation of ideas. However, with the development of artificial intelligence our historical belief that video and audio are reliable records of reality is no longer tenable. Seeing (or hearing) is no longer believing, as the tools available to spread disinformation or misinformation are becoming far more sophisticated.  In this sense, the ‘deep’ that comes to mind is deep dark secret.

If you are interested in pursuing this further, Nina Schick, a former policy adviser for US Democratic presidential candidate Joe Biden, has just published a book – ‘Deep Fakes and the Infocalypse; What You Urgently Need To Know’.

I was thinking about these issues as I immersed myself in the CNN coverage of the Democratic National Convention. Due to the coronavirus pandemic, the convention was (by necessity) virtual. There were many heart-warming and rousing video segments extolling the virtues of the Democratic Nominee – from ordinary citizens to movers and shakers. These all looked real and probably were. However, there was one segment by Sarah Cooper which is a clear example of how deepfaking can be used for political satire. Here she riffs on Donald Trump:

America is clearly struggling with health, economic and social issues. It is clear that at the moment it is stuck between the devil and the deep blue sea. Hopefully, the deep blue will prevail.