Is not the truth the truth?
~Sir John Falstaff
It’s doubtful whether we have really appreciated this Shakespearean conundrum all this while. But Artificial Intelligence (AI) has convinced us, today, that all truths are not truths. For instance, you can create a video of Bin Laden drinking coffee with Barack Obama or one featuring Chinese President Xi Jinping’s face superimposed onto the model Miranda Kerr’s body without people doubting their authenticity. Remember what the former Executive Chairman of Google, Eric Schmidt, said about the internet? “First thing humanity has built which it doesn’t understand”. The internet is the new black magic.
The emergence of deepfake technology is a case in point. Deepfake is a video/audio clip that has been altered with the help of AI. Its power lies in effecting “ultra-realistic alterations to authentic media in a fraction of time, with little cost and without any computer graphic skills”. In 2017, Apple had introduced FaceApp, which could change one’s face, age and gender. Reface, which enables to deepfake a person without a deepfake software or graphics card, followed. Now attempts are afoot at developing a synthetic video deep learning system, which can generate scenes on its own.
Depending on the intention of the impostor, a person may lose or gain reputation. Just one instance from the 2020 US presidential election would prove this. In March 2020, a fake video featured Joe Biden as saying, “We can only re-elect Donald Trump”. Today, every party/country uses this to malign its adversaries. Even the dead are not spared. A recent example is the deepfaking of George Floyd by a Missouri Congressional candidate who circulated a video showing that the real Floyd died in 2016, and that the one who was murdered by the Minneapolis police in 2020 was fake. It was Sophocles who said that the dead had rights. But in the networked world, even the dead are squeezed of all rights and are mere bodies for winning brownie points.
There is also a racial tinge to deepfaking. For instance, FaceApp makes people attractive by lightening their colour and giving them European-like features. The same lacuna is noticed in the facial recognition software used by the police. Criminality is often associated with blackness. The lack of racial diversity in the images used for training the apps accounts for this lapse.
This does not mean that deepfake technology is all negative. It could be used in fields like cinema, art, healthcare, political satire, fashion and education. But as Ovid laments, mankind sees and approves better things, but follows the worse. This is true of deepfake too. It’s used to harass, humiliate, bully, disinform, manipulate and rewrite history. According to a recent poll conducted in the US, 75 percent of the people who saw fake videos believed them to be true! Hence organised disinformation campaign has now become a ‘heavy industry, worldwide’. The Oxford Internet Institute says that 81 countries in the world had waged such campaigns in 2020, and $60 billion has been spent for this.
We will be in a catch-22 situation soon, oscillating between belief and disbelief. This creates a society, which is deficient in trust and co-operation. Many solutions are offered to mitigate these evils. One of these is to ban it as was done by China. But then this will have dire consequences for freedom of expression. Also, such bans would negatively affect technological development.
Mitigation efforts should also focus on detection and education. Developing detection tools and using them to intercept deepfakes and using blockchain timestamping to embed digital signatures in the original media are important here. But no solution could completely prevent technological misuse. Therefore, we need to educate people about the issue. The public should be taught to be sceptical about what they see and hear. It’s wise to follow Bertrand Russell’s will to doubt rather than American psychologist William James’ will to believe. People should also show restraint in uploading images since the probability of deepfaking increases with every image catapulted into the internet.
Former Professor of Political Science, University of Kerala