The politics of ai

The politics of ai

Private capital must be made to pay if it wants to use people’s personal data for commercial gain.

Public discourse around artificial intelligence (AI) is often hijacked by themes that belong in fantasy (an idea with no basis in reality) rather than the real world. Iconic AI from pop culture such as HAL 9000 and Agent Smith epitomize (to be a perfect example or representation of (something)) a Manichaean obsession with the idea of ‘superintelligence’ (‘the Singularity’) that could prove to be good or evil, vested (rights, benefits, or privileges that are typically inherent or legally guaranteed) as it is with the power to turn humans into either immortals or slaves oppressed by parasitic machines. But the Singularity is not what humanity needs to worry about right now.

Machine learning (a more precise term for AI) will certainly continue to surpass human capabilities in specific domains such as medical diagnosis and facial recognition. But an AI that can match human intelligence in all respects is unlikely because it is impossible for AI technology to replicate that which makes human intelligence what it is — its embodiment (someone or something that is a perfect representative or example of a quality, idea) in a biological substrate refined by millions of years of evolutionary feedback loops.

The Big Data leap

This doesn’t mean the advent of AI is no cause for concern. International Data Corporation, the market intelligence agency, estimates that worldwide spending on AI solutions will grow to $57.6 billion by 2021. The lion’s share of the investments is being made by the Big Five: Alphabet, Amazon, Apple, Facebook and Microsoft.

Given the scale of investment that it is attracting from the Big Five, all hailing from the most profitable sector of global capitalism (technology), it’s clear that AI is critical for future profitability. This is in keeping with the dynamic of late capitalism that began in the 1970s.

Capitalism faced a crisis of profitability in the 1970s. Opinions differ regarding its causes, but the global elite had no doubts about the solution: financialisation and globalisation. Also known as the “neo-liberal turn”, it helped solve the problem of falling rates of profit by empowering capital to flow across national borders to wherever the returns were the highest, buy up state-owned assets and enterprises cheaply, and use labour arbitrage (the process of exploiting differences in the price of an asset by simultaneously buying and selling it) to appropriate a greater share of produced value.

The outcome of all this was a diminishing share of wages in profits. So, to prop up (To stop someone or something from falling) demand and keep the economy on the growth path, consumer spending was sustained through debt, which entailed further financialisation of the economy. It was around this time, in the 1990s, that capitalism welcomed its newest saviour: digitalisation.

If financialisation and globalisation made it possible for corporates to tap into markets anywhere in the world, digitalisation gave them the means to do so. Uber is the perfect example of what capitalism wants to be when financialisation, globalisation, and digitalisation come together. Huge volumes of financial capital bankrolled Uber through year after year of huge losses as it expanded across the globe, offering rides at prices that disrupted local transportation markets. But it owned no vehicles, employed no drivers. What it did own was data about customers and commute patterns, and a proprietary algorithm that put them to good use.

The oil is data

AI thus heralds (be a sign that (something) is about to happen) the next phase of digital capitalism where capital accumulation (increase or growth by addition especially when continuous or repeated) is powered by the ‘oil’ of the networked economy: data. To take an obvious example, traditionally the world’s leading content producers, newspapers and television channels, received the bulk of advertising revenue. But in 2017, 25% of global ad revenue and 60% of online advertising were gobbled (to seize upon eagerly) up by two companies that produce no content at all: Facebook and Google.

So where does their value come from? Well, both are platforms: one is a search platform and the other is a social networking platform. As Nick Srnicek argues in Platform Capitalism, businesses structured as platforms are the digital equivalent of oil rigs, ideally placed to mine the networked economy’s most valuable resource by inserting themselves between different sets of users, turning every interaction into a data point, and feeding it all into an algorithm.

From this perspective, India’s own data-mining initiative, the Aadhaar project, is an ambitious attempt to run a single pipeline through multiple oil rigs with the aim of securing free and unlimited access to an endless stream of personal data that could be monetised by whoever controls it.

Platform businesses leverage their ability to scale-up the digitisation of a given activity (Uber digitises taxi rides while Airbnb digitises hospitality) to quickly build monopolies that, in turn, boost their ability to collect more data. Unlike what we’ve heard for a long time about Facebook and Google, this data-collecting spree (an unrestrained indulgence in or outburst of an activity) is not about selling it for money or using it to target advertisements better. Rather, it is about using them to train algorithms. Once a platform is in place to ensure a steady supply of fresh data to train an algorithm, the company can eventually move to a position where it can offer an array of AI solutions for which, unlike online search or social networking, you have to pay.

Need for compensation

All this foregrounds two inter-related issues that citizens must consider carefully: data ownership and labour protection. Put bluntly, the platform-based, chargeable AI services being rolled out by the likes of Amazon and Google were not only made possible by user-generated data, but they often border on rent-seeking. So, there is no reason why people should continue to surrender ownership of their personal data without due compensation.

The time has come to put in place a new data ownership regime so that private capital is made to pay if it wants to use people’s personal data for commercial gain.

Second, AI is set to eliminate thousands of skilled jobs in the services sector — from paralegals and sales executives to drivers and radiologists. Unlike what unfolded in the 20th century when the loss of blue-collar jobs to automation was offset by a boom in service sector employment, the rise of AI isn’t about to open up a great number of jobs in any new sector, which is why tech tycoons such as Elon Musk are advocating a universal basic income.

And yet, every time someone writes a Facebook post or types into Google search, she is not only giving away data about herself, she is also bringing that data into existence, in addition to continually reproducing her physical self so that she can go on being both a producer of data and the agent ensuring that the databases feeding the AI remain populated and active.

So, contrary to neoclassical economic wisdom, AI cannot sever the link between labour and economic value, though it does substitute dead labour (capital) for living labour. A more equitable distribution of the profits derived from data is essential to ensure that the original owner-producers of data get their due share.

Ultimately, the AI-enabled digitalised economy cannot survive without the ‘oil’ that can only come from non-AI (human) sources. Unless citizens exercise political control over how data is mined and used, even without the rise of a ‘superintelligence,’ the bulk of humanity risks being reduced to little more than hyper-connected sheep, kept well-fed and well-entertained on (plat)farms under the supervision of AI well-trained to optimise the production of digital wool.