In this letter exchange, Esmé Partridge and Sarah Kuszynski debate whether AI will be a net positive or negative for our shared future.
Dear Sarah,
I am sure you know, as we all do, the great satisfaction of completing a difficult technical or intellectual task. Whether it is learning a language, writing an essay or practising a craft, the trial and error involved is almost always worth the feeling of accomplishment afterwards. We often take it for granted, but this is a crucial aspect of human fulfilment and flourishing – one that AI could completely undermine.
In the workplace, automation is already making hundreds of roles redundant. The World Economic Forum predicts that AI will replace 85 million jobs by 2025 and, in their most recent Future of Jobs report, found that 25% of companies expect it to cause redundancies. Outside the domain of employment, the tools offered in particular by generative AI are increasingly removing the need for manual input in all areas of life. Content can be written by ChatGPT; graphics can be conjured up by MidJourney; and our lives themselves can be organised by AI schedulers like Motion.
While we might think that the radical cutting of corners offered by these technologies will improve our quality of life, the reality is quite the opposite. Not only will employees suffer as a result of companies deeming AI more cost efficient than real people, but the increasing reliance on automation may ultimately pose a threat to human cognition and creativity. What you do not use, you lose: the more dependent we are on technology, the weaker our own capacities could become.
This is a concern not only because it puts us in danger if, for whatever reason, the technology fails; it also, in removing human agency from the creative process, diminishes our experience of life. We may no longer have to endure menial tasks, but we will also cease to experience the satisfaction that comes from discipline and self-improvement. Central to Aristotle’s idea of human flourishing, or eudaimonia, is the pursuit of technical skills so that we may each actualise our unique potential. It is the very pursuit of work itself – and not just the final product – that helps fulfil our ultimate purpose, or telos. Since AI removes the need to hone such skills, it inhibits this fulfilment and thus our flourishing overall.
This is why leftist fantasies like “fully automated luxury communism” are misguided: while automation might seem to offer a better way of life for ordinary people – by way of supposedly liberating them from the demands of capitalism – it is far more likely to rob them of agency and purpose.
The technocratic tendency of Silicon Valley is dangerous for this reason. Since AI can only be configured by a select few ‘experts’ at the top, ordinary people are no longer able to participate in the technical process.
This not only diminishes their quality of life at a personal level, but also puts the nation in danger in times of geopolitical instability. Instead of relying on an invisible class of ‘experts’ to manage our lives, it is surely better to invest in cultivating the skills and talents of real people.
Dear Esmé,
I certainly agree that we gain immense amounts of satisfaction from the successes and failures of learning, as it is through these experiences that we grow and develop as individuals.
However, to say that AI will completely undermine human cognition and creativity risks portraying AI as some ‘monolithic demiurge’ which should either be entirely resisted or, due to its transformative capabilities, completely accepted. Instead, we need to be looking for ways to incorporate AI’s useful aspects – those that complement and could even unleash further creativity, through its productivity enhancing and time-saving potential – and reigning in the harmful. Generative AI could raise overall productivity levels by 18% over the next ten years, as well as raise global GDP by around 7%.
Presently, we appear to be in the early stages of an AI revolution the full impacts of which are as yet hard to tell. So far, generative AI tools only seem to threaten mind-numbing, bureaucratic work – I am sure we can agree that bureaucracy has spiralled out of control in recent years, with endless form filling, email drafting and meeting scheduling. As such, it has reduced the quality time we spend on tasks which afford us opportunities for genuine, meaningful self-improvement. When the time saved by using generative AI to do these tasks could be spent on more complex, intellectually-challenging work, why choose not to engage with it?
Further, while you cite Aristotle’s belief that happiness derives from our sense of self-accomplishment, and view AI as an impediment to this, one Karl Marx similarly feared that increasing mechanisation would rob workers of the opportunity to achieve fulfilment — a dynamic which he believed would lead to a global proletariat revolution. As this remains to be seen perhaps we should be less fearful that AI-powered ‘mechanisation’ will prevent self-actualisation.
What is more, AI need not threaten the creative process. First, AI, as it stands, presents only the illusion of creativity: AI-produced paintings, music and literature are simply built from the data that is used to train it. Thus, true creative power and genuinely new ideas will only remain achievable through human thought and interaction.
Second, AI-based art could become its own distinct sub-field, indeed AI-enabled art is already giving rise to dedicated digital art groups. In this sense, human creativity is being diversified rather than purely undermined. Although the boundaries between human; AI and co-produced art should be maintained, it is nonetheless clear that AI is becoming just one of the many options artists have to prompt their work.
Beyond the art world, at the frontiers of creative problem solving, AI is also helping us to address previously unsolvable mathematical and medical issues. Such challenges are so complex and multifaceted that, even with the best minds focused on them, they often remain unsolved. Yet, with the assistance of AI, new mathematical proofs and medical treatments are already being found. In these cases, the initial creative process is primarily human-driven, with the humans involved still requiring in-depth subject knowledge and complex technical skills – the AI complements rather than overrides human creativity, technical expertise and purpose.
That said, AI will undoubtedly replace jobs and result in redundancies. Machines have been substituting human workers since the first industrial revolution, but each major advancement has also prompted labour market transformations. After all, 60% of workers are employed in jobs that did not exist in 1940. This is creative destruction at work, where old methods make way for the new. Indeed, jobs such as data curators and ethics and governance specialists will be brought to prominence by AI. The threat of AI to employment can be, therefore, overstated.
Humans have an innate capacity for change. As Tom Chatfield says in his recent book, Wise Animals, we “are self-made and self-remaking: inheritors and custodians of a contingent world.” This does not however mean we should disregard caution and wholeheartedly embrace the transhumanist, accelerationism expounded by Silicon Valley elites. With a technology as complex and heterogeneous as AI, there is a need for a healthy dose of realism to ensure that we pause and think, something which the dominant mindset in Silicon Valley actively discourages. Therefore, we need a pragmatic, human-centric approach to AI so that it complements rather than undermines our creativity.
You make the important point that AI is not a monolith, which is true inasmuch as the technology itself is inherently neither ‘good’ nor ‘bad.’ At present, AI only reproduces and reconfigures data that we put into it. As such, it can be programmed to serve ethical pursuits as much as it can unethical ones. It is also true that there is potential for AI to be applied in such a way that works with, rather than against, our human faculties – where this is actualised, the dangers discussed in my initial letter can indeed be mitigated. One example of this would be language learning apps like Duolingo, which now use AI to generate exercises tailored to each learner based on their previous mistakes. Here, the technology actually serves to improve human memory and skill – so, as you suggest, there is potential for it to be used in a more humanistic way.
The question, though, is how likely it is that this approach will prevail. I am not convinced that the majority of those currently taking advantage of AI are really interested in our self-betterment. Take X (formerly Twitter) since Musk’s takeover, as an example: whereas once users would see posts mainly from accounts they chose to follow, the algorithm now feeds us content that is calculated to provoke outrage and polarisation. This is because this generates more engagement, and so more advertisement revenue. Unfortunately, encouraging people to indulge in base behaviours – be it tribalism or simply mindless consumerism – tends to be more profitable than self-improvement. It is no surprise, then, that AI is largely being used for these questionable ends.
If our politicians had a robust conception of the common good – and were willing to uphold it in the public square – we would be better able to ensure that AI is used for the genuine improvement of society. Where social liberalism and free market capitalism are indifferent towards the good, however, intervention is unlikely. This poses a particular problem when it comes to technology that has the power to artificially exacerbate the very worst aspects of human nature and spread pernicious ideas beyond our control. It is arguably because of unregulated social media, for example, that an entire generation of young people have been exposed to eating disorders and self-harm.
This is primarily my concern: AI may well have the potential to be used for good and to improve society, but how do we ensure that this is actualised? Left to our own devices, human beings – and, in turn, the market – are easily corrupted by greed and power. This, of course, raises broader questions about human nature and the extent to which governments should intervene. Honing in on the issue of AI, though, the fact that it has an unprecedented potential to ramify already existing tendencies in society – both good and bad – should at least make us consider the possibility of regulation.
Historically, when faced with radical technological change, societies have tended to impose some kind of measure in response – even if only temporarily, so that they can assess what they are dealing with. For example, the rise of the printing press in the fifteenth century was followed by the Catholic Church’s Index of Forbidden Books. Although we might look back on this as an authoritarian and repressive move, the Index helped limit the pace of social transformation and gave cultures and governments time to adapt. Such precautions have an important role to play in giving us the breathing space to understand the implications of innovation and to craft the laws and policies that curb destructive excesses.
Even free markets must confront the problem of externalities. Those who develop and profit from AI must consider the potential costs of their innovation on wider society. This is not an oppressive move, but one that recognises that some powers are too great to risk letting them spiral out of control. It is for this reason I believe that, if we allow AI to proliferate, some regulation must be implemented – otherwise there is no guarantee that its good uses will triumph over the bad.
Dear Esmé,
Humans, of course, both shape and are shaped by the technologies we create, and just as there are both good and bad people, so can AI be harnessed for nefarious and well-meaning purposes. As such, who sets the intentions, values and codes of ethics of AI systems matters greatly for the future direction of AI. It is hard to prevent AI systems from being monetised by tech giants, for whom profit-maximisation is, naturally, one of their goals. I do not disagree that we must therefore enable a cultural and regulatory environment where the potential for AI to be used for individual and public betterment is actively encouraged.
To ensure this, a healthy public square is necessary, but what is more urgently required is civic education and public awareness campaigns surrounding AI – both its benefits and its dangers. Civil society organisations, in particular, should be empowered to provide people with the knowledge and tools to discern for themselves what is harmful or base and what will enable self-improvement. Reassuringly, we are already seeing something like this in campaigns that aim to help people identify mis-and-disinformation so that it is not permitted to poison public discourse.
Undoubtedly, realising the above will require a degree of government intervention, however, if the potential harms of AI-powered social media algorithms are simply grouped together with other public health concerns, from viral pandemics to obesity, we may elicit overbearing and ill-thought-through responses, which could prove more damaging than the harms the intervention was intended to prevent. For example, we are only now coming to understand the damage that the COVID-19 lockdowns had on young children’s development and educational progress. Such policies were intended to be in the people’s best interests, yet resulted in additional societal harms. In the case of AI, Google’s generative AI tool, Gemini, exemplifies my concern. When prompted to create images of mediaeval popes or kings it created historically inaccurate figures – incorrectly displaying female figures. Google, in trying to regulate the AI and well-meaningly promote inclusiveness, inadvertently risked distorting truth and reality.
Returning to my previous example, lockdowns also gave rise to a worrying orthodoxy of thought and government control over daily life. I fear that overly restricting what AI-powered algorithms and chatbots show or say to us could implicitly shape how we think, giving rise to only one conception of what is right and proper for people to think . What is more, given that control over algorithms and chatbots lie with Silicon Valley elites, it is likely the values and ethics of these companies are the ones that would become the new orthodoxy.
Social media must remain a marketplace of ideas, so providing people with an education that places value on freedom of speech, thought and expression is more vital than ever. After all, it is only through exposure in a public square that pernicious ideas can be refuted. On top of this, overzealous regulation often has to be rolled back due to its ineffectuality and impracticality. For example, Italy banned ChatGPT when it first emerged, lifting the ban only a few months later. Interventions that try to ignore rather than engage with technological advances are clearly futile endeavours. To both improve the efficacy of regulation and preserve the marketplace of ideas, interventions must thus only ever be cautiously made.
We should not downplay the potential harms of social media use, especially for adolescents, yet we also need to ensure that we foster a culture of responsibility rather than one that coddles and overly restricts our young people. We cannot allow AI regulation to further inflate belief in the ability of government to solve problems which are, at their heart, problems of individual responsibility. As such, regulation is important, but getting the approach to it right is more so. At present, the UK’s lighter touch approach exemplified by the Government’s recent White Paper on AI holds promise – it seeks to avoid quashing innovation while, at the same time, trying to limit AI’s most concerning risks.
So, while I agree that we need a regulatory safety net, regulation alone cannot be the solution. A drive for civic education and promoting a culture of individual responsibility is what is increasingly required. This will help realise the potential for AI to enhance, rather than hinder, human flourishing.
Esmé Partridge is a regular contributor to various publications, including the Critic and UnHerd. Sarah Kuszynski is a Senior Researcher at Bright Blue.
This exchange was written for the latest edition of Centre Write.
Image credits: julien Tromeur
Read more from our June 2024 Centre Write magazine, ‘Generation AI?’ here.