This Blog is re-posted from John Howarth’s MEP website and based on his contribution to “An Industry Vision for a Renewed Europe”, a debate organised at the European Parliament by the European Forum for Manufacturing on 6 March 2019.
We are told, and it happens to be true, that we stand at a time of vast technological change that will transform human lives in a future beyond our imaginations. It is true because it has always been true – at any time in the past 300 years we could have said much the same thing with confidence. The industrial and technological revolutions have been a time of permanent and most frequently accelerating revolution. And while war has from time to time slowed down development it has also been the dark catalyst of technological progress.
The question is not whether humanity chooses digital transformation but how we can learn from previous transformations in the sure and certain knowledge that the likely pattern is one that we have seen before. It is just another change: every technological leap has produced vast swathes of wealth but with greatly unequal distribution and unequal markets prove unsustainable; every process automation has removed turgid tasks from the workplace but has also removed human skills from processes.
Technology always drives change faster than legislators and social policy makers can regulate the parameters and consequences. We know that the greatest innovations will drive markets in particular directions. The challenges are: how do we retain sustainability, how do we ensure that markets retain humanity, how do we ensure that people genuinely benefit from these great leaps forward.
Robotics and automation have overwhelmingly replaced tasks that required low to no skills but as robotic automation becomes more sophisticated tasks requiring greater dexterity have been replaced. Artificial intelligence, most fundamentally, introduces decision making to automation. Decision making applies certain criteria to processes – and they can be quite complex, considering intricate conditionalities and building in judgements that can be seen as protecting the individual consumer as well as the provider corporation. Experience of the pseudo-judicial functions of public administration would, however, suggest that some judgements and not best handled by binary systems because in the end they are not always binary.
So assuming we acknowledge as policy makers both the productive bonus of automation and its dystopian tendencies and how these deliver change in a market economy we can being to provide answers to the direction in which society needs to travel for the social element of our economy to deliver societies at ease with themselves.
The ability to engage in a digital world
The real danger is of a digital world in which substantial minorities within otherwise wealthy societies are excluded from the productive economy. This divide is not about class, background or income in the first instance but about the education and skills that enable individuals to engage. An economy that is fully able to take advantage of the wealth creating opportunities of the digital world will rely as never before on a technologically literate workforce.
Likewise, our public sphere, depending as it does for its legitimacy on participation and democratic engagement, requires a broadly based, representative, digitally enabled population. Both require different thinking on the outputs of education. The nature of the evolving economy will require an education and skills system that thinks (metaphorically) in terms of maintenance as well as manufacture. We need no longer to think of re-skilling for transition on an industry-by-industry basis but in terms of constant and ongoing skills development across the workforce. This should become a central element in EU and Member State funding of programmes for ‘Just Transition’ during the 2021-27 MFF and beyond.
Algorithms are not neutral
Few, after the Cambridge Analytica affair would contend that an algorithm can never reflect bias. However the industry that produces algorithms – and therefore develops core aspects of artificial intelligence – leans heavily toward a particular type. The technologist caricatured in popular culture only chimes comedically because they are true to the experiences of the audience. The software engineering sector was white, male, middle class, college educated and overwhelmingly 25-40 years of age. This is less true than it once was, or at least the first of those, however the point is this.
If algorithms reflect the societal biases of their authors then they will produce outcomes that reflect those biases. Data sets with inbuilt bias will also produce skewed outcomes. In the automation of HR/recruitment, student applications and insurance processing the implications are obvious, though even in mechanical automation there is early research to suggest that being treat equally by the machine cannot be taken for granted. Further research will be needed to determine if the early suggestion that a black woman is more likely to be involved in a collision with a driverless vehicle than a white man (according to a study by the Georgia Institute of Technology). These fears may prove to be false – but how will we know unless outcomes are critically examined and unless access to the data being used by automated processes is available to independent researchers?
Where the greatest gains will lie
The more productive wealth generation will take place in the areas of technology that make possible the things that before were impossible or limited by the physical and political world. They will create the next billionaires and produce the next social exclusions. It will happen because these areas are always the most difficult for legislators to keep up – and some may prove to be out of political reach. Blockchain technology is receiving all sorts of attention for exactly this reason. This is not to say that the automation of existing processes will not be productive, just not exponentially so.
What does this mean?
Digitisation is happening. It cannot, should not and will not be stopped or regulated out of existence because markets will find a way. Luddism doesn’t work – it will only increase the digital divide. Just as the challenge of China cannot be successfully addressed by protectionism but only by Europe becoming an effective inward investor and stakeholder in Chinese domestic, international and digital markets so digitisation will only become socially inclusive by social and public actors working with digital innovators.
The long term success of Europe’s digital/AI economy requires the broadest possible section of consumers who can afford to buy the products of that economy. It requires an economy producing products to meet the needs of a diverse society. It requires a digital economy not confined to a monoculture, even a monoculture of the majority but which enhances cultural diversity and harnesses that diversity to create better products for more sustainable markets.