AI is challenging us to rethink the distinction between ‘real’ and ‘fake’

By allowing that authenticity and artistic merit can be achieved through diverse means, including the integration of synthetic elements or AI help, we can reimagine the boundaries of creativity and originality
But this paradigm shift comes with risks surrounding the discerning of truth from deception

Allen Day

Published: 3:30am, 24 Jun 2023

Why you can trust SCMP

As we immerse ourselves in the digital world, we enable artificial intelligence (AI) to learn from us and it is rapidly becoming more able to emulate human behaviour. By transforming the way we communicate with one another and engage with information, the symbiosis between analogue and digital realms will change our individual and social fabric.

The social sciences use a concept called authenticity to discriminate between original and derived works. As more of our assets are partially or wholly digital, authenticity is increasingly hard to benchmark. It may be that authentic/inauthentic has always been a false dichotomy, that there never was any clear criteria for discrimination between the two, and that it was merely a matter of what was written down and could be referenced.

Regardless, much of our society is underpinned by intellectual property rights, which depend upon authenticity. Thus, as creative industries adopt AI for commercial applications, we reach a crisis: a need to rebalance the benefits of permissive, fair use against copyright.

Earlier this year, The Hague’s Mauritshuis museum replaced Vermeer’s iconic Girl with a Pearl Earring with an AI-generated artwork inspired by the original. Created by Berlin-based artist Julian van Dieken using Midjourney, it fuelled a heated debate around artist agency and copyright issues.

Easy access to image generators such as Stable Diffusion and Dall-E have democratised collaborative creation with AI. At the same time, the impressive capabilities of text-generating large language models are being explored, leading to a boom in so-called prompt engineering.

AI’s impact on the creative process has entered our everyday lives with unprecedented speed. Within two months of its launch, ChatGPT garnered 100 million users, making the AI chatbot the fastest-growing consumer app ever.

A visitor takes a picture of an image designed with artificial intelligence by Berlin-based digital creator Julian van Dieken, inspired by Johannes Vermeer’s painting “Girl with a Pearl Earring” at the Mauritshuis museum in The Hague on March 9. Photo: AFP

What the text and image-generating AI models have in common is that they are constructed, or “trained”, by analysing large collections of digital images, videos, text and audio.

Creative industries affected by the transformative power of generative AI seek a balance between innovation, protection of intellectual property and the preservation of artistic integrity.

In April, the music industry was shaken by an AI-generated song which simulated the voices of Drake and The Weeknd. After the song was taken down from Spotify and other streaming services, Universal Music Group released a statement criticising the rise of “deep fakes”.

The music label said using its artists’ music to train generative AI “represents both a breach of our agreements and a violation of copyright law”. As new AI-generated hits continued to pop up, major artists joined the conversation and industry insiders explored questions such as whether music recorded by humans could become the exception rather than the rule.

With AI integrated into the creative process, traditional and emerging notions of authorship and ownership are at the nexus of sociocultural, economic and regulatory debates. This issue is increasingly urgent given the rapid advancement and refinement of the models. Lawsuits claiming unauthorised use of copyrighted material as training data are piling up.

But global consensus remains elusive. In Japan, for instance, the training of AI models using publicly available data is considered legally permissible, said Keiko Nagaoka, Minister for Education, Culture, Sports, Science and Technology.

02:22

ChatGPT being tried out by Yokosuka city government workers after OpenAI founder visits Japan

AI’s advancing capabilities, which promise lower costs, faster delivery and a superior product, have brought synthetic media into the spotlight. While creating a significant market opportunity, this development also raises concerns about the erosion of authenticity.

To examine this, we must acknowledge the orthogonal nature of real vs fake and organic vs synthetic, and re-evaluate our conventional understanding of “original work”. The notion must encompass the possibilities offered by innovative tools.

By allowing that authenticity and artistic merit can be achieved through diverse means, including the integration of synthetic elements or AI help, we can reimagine the boundaries of creativity and originality.

02:36

AI-generated image wins photography award, but artist turns it down

But this paradigm shift comes with risks surrounding the discerning of truth from deception. The widespread adoption of synthetic media calls for heightened vigilance and critical thinking. Most recently, the European Union urged social media platforms to “immediately” start labelling AI-generated content as part of moves to combat fake news and disinformation.

The “fake” Drake track is a reference point for observing how our hunger for bespoke content plays out in the evolving landscape of digital creation. Given the vast depths of data lakes of our individual preferences, habits and interests, the concept of hyper-personalisation beckons.

Streaming services already tailor recommendations, leveraging data like viewing habits and search history. Social media platforms employ algorithms that sift through content, delivering only “relevant” posts to increase engagement.

02:08

Virtual hosts are rising stars of China’s online shopping platforms

Video game developers are designing increasingly immersive experiences. AI tools not only allow marketers to analyse a wealth of customer data, they can now create next-generation customer touchpoints with personalised content.

This next-level customisation goes beyond deeply resonating with a target audience, and appeals on a completely personal level. By leveraging AI algorithms, people can enjoy content that aligns with their distinct interests in music, film, literature or other media forms.

The content becomes a reflection of their individuality, resulting in a more engaging experience – even as content production costs decline.

But the convergence of AI-driven personalisation and our digital immersion risks narrowing our exposure to diverse perspectives and shared cultural experiences.

Such curated versions of reality could hinder serendipitous discovery and exposure to alternative viewpoints – creating an isolation where the distinction between fake and real becomes highly subjective, placing the responsibility of discernment solely on personal judgment.

Allen Day is a Google executive and spokesperson, developer advocate and creator of Google Cloud’s cryptocurrency public data sets

Post