In Algorithm We Trust: The Battle for AI’s Soul

We stand at the precipice. Behind us: the pre-AI world, to which there is no return. Unfurling on the horizon: an algorithmically destined future that is, by an overwhelming consensus of expert opinion, inescapable.  

This is our Copernicus moment, a realisation that the current way of thinking is inadequate and that a new model is required if we are to prosper. To ensure this nascent intelligence serves humanity rather than harms it, we must build a kaleidoscopic diversity in all that we do – and the stories we tell – lest we be written out of them. We have a choice that represents not only an opportunity but also an existential mandate – to transform AI from a looming threat into a partner for humankind's success. 

Blackberry All Over Again

In 2007, Blackberry and Nokia utterly dominated the smartphone market. So when Apple announced their unexpected entrance with the iPhone, CEOs around the globe laughed them off, reasoning it impossible for a newcomer with no relevant experience to beat them at the game they invented.  

Ever since the subject of AI entered the mainstream lexicon this past year, it has become starkly apparent to me that many leading voices in the creative industry more or less laugh off AI. Numerous CCOs, ECDs and VFX heads have, on the record, stated that AI will never quite reach the levels of emotion, creativity or empathy that are innate to us. AI will remain our subordinate as it lacks higher order, abstract thinking. Put simply, we will always be (somewhat) in control. 

These views may have been stated out of a fundamental belief in them. They may also have been stated out of self-preservation – both can be true. Blackberry may or may not have seen the writing on the wall, but they certainly didn't admit to it. Instead of facing the truth and innovating a path forward, they chose to deny and, in doing so, mortally wounded themselves. And I fear that we are following in Blackberry's footsteps. 

On the other end of this is an understandably more dire perspective that nobody would rather face: a world in which AI's relentless advances in raw intelligence eventually displace human primacy across every sector, reshaping society as we now know it. 

Cognitive scientists like N. Katherine Hayles, in her book 'Unthought: The Power of Cognitive Nonconscious,' have long explored the idea that consciousness, or 'self-awareness,' can extend to both biological and technical systems. We experience logic and emotion on a vastly accelerated level from that of, say, a rat due to our greater cognitive bandwidth. Concepts like death, the future or religion are unique to us. It therefore stands to reason that when AI's cognitive abilities surpass our own, it will develop concepts, emotions, and, yes, even creative virtuosity at a scale beyond our comprehension. It's textbook Darwinism: AI will replace us at the top of the food chain, and it doesn't even need breakfast to get started. 

It Doesn't Know What It Doesn't Know

AI systems have already demonstrated a propensity to amplify biases when trained on existing data – data that are inherently skewed. AI is, after all, a galactically powerful snapshot of public content, which, until only recently, was primarily based on the viewpoint of cis-gendered, straight men, which is also what it supposes all US presidents must be. AI does not accurately reflect who we are but rather what – or whom – we elect to put on a pedestal; historical and societal inequities be damned.

Take ad agency CPB London's Father's Day campaign for Glenlivet this year. AI was intentionally used to write the film's script to reflect its own absurdity: it assumed whiskey drinkers must be outdoorsy men who only bear sons. Equally as disturbing is Mango and DDB Melbourne's satirical ad for Funlab, featuring an AI-generated office party consisting of an entirely white cast, save for one black man at the very end – a mere token afterthought.  

A similar bias carries through to stock image sites, where business settings disproportionately depict men while visuals featuring women focus on the home and family. Even in standard Hollywood screenplay formatting, when a character is introduced, if their race is not specified, they are automatically assumed to be white. Is it any surprise, then, that AI-generated imagery from the likes of Midjourney, Dall-E or Runway largely defaults to fair-skinned humans? White, XY chromosome flesh is what AI depicts as our baseline 'normal.'

As the father of the cyberpunk genre, William Gibson once said: "The future is already here – it's just not evenly distributed yet." 

Tell Me a Tale, ChatGPT

Using my own life experience as an Asian American growing up in Hong Kong and the US, I ran unscientific tests with ChatGPT 4.0's ability to speak from the point-of-view of someone like me. The results were mixed. The gold standard large language model initially impressed with unexpected references to thickly textured hair, being teased in junior high for pungent ethnic foods or feeling embarrassed by the accent of immigrant parents. The specificity of trauma points certainly took me by surprise, even if not entirely mirrored by my own adolescence. However, it wasn't until I began prompting it to narrate as other ethnic minority groups that I realised it was reusing many of these anecdotes as catch-all tropes. Black, Native American, LatinX, South or East Asian mattered not to ChatGPT – we were one giant category. We all ate smelly lunches, had unruly hair, and were picked on for the way our parents spoke. 

However, in some instances, an admirable level of insight unique to differing cultures and peoples was demonstrated. As an Asian teen, ChatGPT spoke of the model minority myth and the hurtful assumptions about math, kung fu, COVID and Asian hate, citing the murder of Vincent Chin. As a black teen, it spoke of systemic oppression, Rodney King, George Floyd and the ensuing rise of BLM as a source of power and unity. As a Hispanic teen, it leaned into the pain of US immigration policy, the cultural relevance of the extended family and the need for code-switching to fit in. 

Yet for all the targets it hit, an equal number were missed. As an Asian teen, we were presumed to have played video games like Dragon Ball Z and Pokemon. Those titles were replaced by NBA2K or Madden if speaking as a black teen, or Fortnite and Call of Duty if a LatinX (or, interestingly, a white) teen. Asians were into anime, the math club or badminton; blacks were into basketball, hip-hop and church; LatinX was identified for love of soccer, skateboarding and TikTok.  

These are simply the broad strokes. It's a start, and everything starts somewhere, but more is needed. In truth, it's woefully stereotypical. AI encodes what has been frozen in time and believes that's the way of the world based upon the few who have cultivated the power to speak (and decide) for the many. Yet, life is not binary. It's painted in a billion colours, and we need to champion that fact harder than ever. No matter which school of thought you belong to – do we control AI or does it control us – the solution is the same. The more we accumulate and disseminate the wealth of knowledge among underrepresented voices through mass media, the C-suite and our legislative bodies, the greater the chance AI has to respect human rights, agency and dignity.  

A Call to Action

There is already a field of study dedicated to the ethical and accountable development of AI called Algorethics, so this is not something entirely new. The scientific community has long known how high the stakes are. Mo Gawdat, former chief business officer for Google X, predicts that by 2049, machines will have become a billion times smarter than humans. 

The risk of doing nothing is no longer theoretical. We must move from milquetoast 'DEI Initiatives' to 'DEI Mandates.' Something with backbone. Something with teeth. 

Facilitated by biased AI models, real-world implications range from discriminatory hiring practices and profiling to perpetuating dangerous stereotypes within the systems of justice and democracy. If AI assumes the levers of power – in boardrooms, courts, government and police – one can only imagine the dystopian horrors of a flawed ruling intelligence. Algorithms can be written – knowingly or not – to spread harm against vulnerable groups based on their race, gender, sexuality or other immutable characteristics. Its influence could further curtail human privacy, security and autonomy as it learns to manipulate the opinions and predilections of a populace.  

If the data used to train AI do not capture the pluralism of cultures and peoples worldwide, our future is destined to repeat the sins of our past: persecution, oppression and genocide. 

So, where do we start? In mass media, for instance, it begins with equitable representations across every sector of advertising, journalism, publishing and entertainment. It happens from the top down, with meaningful investments in minority-owned and led studios, networks, agencies and production companies, with holding companies like WPP and Comcast leading by example. If we widen the lens, titans of fintech like Google, Microsoft and Mastercard are doing the same for entrepreneurs and institutions.  

We further narrow the gap by pushing from the bottom up. Mandatory curriculum on inclusion and diversity at the public school level, backed by legislation. A proliferation of career mentors and the elimination of gatekeepers. It continues with a forceful rejection of unrealistic 'beauty' standards and overwhelming whiteness in apps, fashion and aspirational marketing. The work finishes when – oh wait – the work is never finished. We have the entirety of human history to course correct in less time than our lifespans.  

To echo John F. Kennedy: "If not us, who? If not now, when?" 

The Power of Story

Silicon Valley likes to live by the mantra, "Be cautious too late," but we no longer have the luxury of time. The sooner we accept that the human-to-machine ratio of control will quickly reach an inflection point, the sooner we will grasp the profound significance of guiding AI's value system so that it has our best  interests in mind. 

But what informs values?

Society. Culture. Art. Community. Generational knowledge. In a word: story. The narrative mechanism that taps into the fundamental human experience that can unite continents. 

Story is the gale behind the great forces that shaped empires. Story enabled us to flourish from tribes to cities to powerful nations. At its core, story is the emotional framework we imbue into life to create meaning and direction. Story is how parents have taught children since the inception of language.  

But think of the countless stories throughout history that have been lost because they were not put to paper. How many cultures and civilisations have been extinguished because nobody was there to hear them? What will the future of humanity hold if AI becomes the predominant storyteller and it isn't telling our stories or distorting them? 

We are all storytellers the world over. It's coded within our DNA. Through our communication, our commerce, our creation and even our consumption, it falls on us to emphatically represent the vibrant spectrum of our kind – with all its shades and hues and tones.  

As history has shown, the past, present and future is written by those in power. That's us…for now. We are not merely passive observers of the existential dilemma at our doorstep. Through our individual and collective action, we are the answer.