The public deserves socially beneficial technology in return for its AI investment

Tech deregulation

Recent years have been marked by strong campaigns to deregulate the tech sector to “cut the red tape” and unleash the full potential of AI. The narrative of having to win “the AI race” and establish digital sovereignty in light of a difficult geopolitical situation has fuelled a wave of deregulation. This lopsided focus on economic growth has led to questionable policy decisions.

It has become almost sacrilegious to question the policy goal of growing the AI industry. There is an underlying assumption that AI growth will necessarily be beneficial for everyone. AI companies are often portrayed as “heroes” that will improve people’s lives, if we will just let them innovate free and uninhibited.  

But is that really the case?

In this situation we can take lessons from a classic Greek myth – the tale of Perseus and Medusa. Perseus, son of divine parentage, is said to be a strong and fearless hero who journeys to slay Medusa, a seemingly invincible monster whose mere gaze can turn living beings into stone.

Yet, as history writer Natalie Haynes points out, Perseus did not work alone. He received significant aid from the gods to help him in his quest, including a cap of invisibility from Hades, winged shoes from Hermes, a sword from Hephaestus, and a shield, enchanted bag, and strategic council from Athena. Without this aid, Perseus could not have slain Medusa. Nonetheless, he is commonly thought of as a “self-made” hero.

Today, similar narratives are crafted about AI companies and their founders. These “self-made” innovators are said to make the impossible, possible. However, just like Perseus, they rely on substantial public assistance from the state to realise their goals.

The public pays

The public already “pays” for AI through a wide range of substantial (in)direct subsidies. The aforementioned efforts to deregulate and weaken data protection, labour, environmental, liability, non-discrimination, and intellectual property laws are a key form of regulatory subsidy, but equally important are the tax breaks, direct public funding, and privileged access to public lands, water, electricity, data, and infrastructure companies already receive. Policymakers often take decisions to offer this help without meaningfully consulting the people ultimately paying for it.

AI companies would not be successful without the public’s support. Our data, our land, our electricity, our water and our weakened fundamental and individual rights all help support the AI industry

The EU’s Digital Omnibus and the Cloud and AI Development Act (CADA), among many other political initiatives, are chief examples of the attempt to weaken regulatory restrictions on technology at the cost of fundamental rights and environmental protections, and pave the way for further roll-out of general-purpose AI products and services, as well as the datacentres needed to power them.

The Omnibus and CADA are driven by a desire to unleash the “full potential” of AI and improve competitiveness. Deregulation is allegedly a necessary cost to achieve these goals.  

The problem with this argument for deregulation for the sake of innovation is that relevant counterfactuals are never considered. For example, what are the costs of deregulation? How many people will lose access to legal recourse when AI harms them?  How many more people will face AI-driven discrimination due to weakened equality standards? How many people will have their health harmed by weakened environmental laws, or face restricted access to essential public resources?

Unfulfilled promises

These are not hypothetical questions – the public is already facing real costs from state support of the AI industry. For example, further building out of AI datacentres is already underway in Spain, Germany, the UK, Italy, the Netherlands, France and Ireland.

Reports in Ireland show that demand for electricity and water already outstrips available supply, meaning residents are facing increased prices for essential utilities. Residents are also losing access to public lands. Historical trends suggest that promises that new datacentres will create significant employment opportunities are rarely ever fully fulfilled.

Through the various forms of support and subsidies being paid, the public is in all but name a shareholder of AI companies. And yet, they are promised little in return to justify the subsidies granted by policymakers on their behalf.

We need to start asking ourselves different questions about the AI industry. Rather than naively assuming all AI innovation is socially beneficial, we should ask: what exactly does the public receive in return beyond access to a technology of dubious value? Why are AI companies the only beneficiaries of current arrangements, while the public only pays the costs?

AI systems pose severe and uncertain risks for society and yet are sold predominantly on the basis that they enable economic growth and competitiveness, with little attention given to social or non-financial benefits. Economic growth is no guarantee of economic equality, and yet many of the risks and costs of AI are socialised.

Optimising for happiness

I find it tragic that the current public narrative pits economic growth, competitiveness, and digital sovereignty against fundamental and individual rights. However, profitability and protecting human rights do not have to be in conflict.

In my new paper Optimising for happiness in EU technology policy to repay public subsidies for AI, we show how to take a different pathway in public support for AI development. We call to rethink AI policy to ensure the industry actually delivers socially beneficial products and services.

Key to our reconceptualisation is the idea that gross domestic product (GDP) alone is an insufficient measure of social value and progress. Additional policy metrics, such as happiness and wellbeing, are needed to ensure the public actually benefits from the technology it subsidises.

We are not alone in our efforts to re-orient policymaking towards more holistic measures of social value. Nobel Prize laureates Joseph Stiglitz and Amartya Sen, as well as the renowned economist Jean Paul Fitoussi, have long criticised the “limits of GDP as an indicator of economic performance and social progress.”

We are standing at a crucial juncture where backlash against the AI industry and frustration with public policy is understandably growing globally.

Socially beneficial by design

I still strongly believe that AI can be a force for good. But AI is not socially beneficial by default – it must be consciously designed to be socially beneficial. The AI industry’s guiding principle should be to create a technology that demonstrably increases people’s happiness and wellbeing, drawing on decades of research from the cognitive sciences about these concepts.

Policymakers should likewise incorporate this principle into decisions about future support for the industry. Specifically, when policymakers are deciding whether subsidies should be granted or extended for the AI industry, they should require companies to predict measurable benefits for public happiness and hold them accountable for delivering on them.

In Greek myth, Perseus was smart enough to not take his divine benefactors for granted. He eventually returned all the gifts that were bestowed on him. He paid particular tribute to Athena by giving her the head of Medusa which the goddess now proudly wears on her chest armour.

We should start expecting AI companies to do the same. It is important to understand that AI companies would not be successful without the public’s support. Our data, our land, our electricity, our water and our weakened fundamental and individual rights all help support the AI industry.

It is past time that the public receives a technology that is truly socially beneficial, and not simply economically valuable, in return for its investment.