Digital technologies have yet to earn their ethical spurs

28 January 2019 | Onora O'Neill

A new, a vast, and a powerful language

In words that are at one and the same time prophetic, optimistic and haunting Ada Lovelace wrote:

“A new, a vast, and a powerful language is developed for the future use of analysis, in which to yield its truths so that these may become of more speedy and accurate practical application for the purposes of mankind than the means hitherto in  our possession have rendered possible.”

She saw beyond the possibility of a computational engine that could make numerical calculations to something of wider and deeper import. And now the realisation of a version – or perhaps many versions – of that vision seems close at hand: but there is a long way to go before we are clear how the practical application of digital technologies can best support “the purposes of mankind”.  I suspect that this reflects some quite disparate issues which the Ada Lovelace Institute will be well placed to address, and I will mention just a few of them.

Innovation vs Ethics?

As it seems to me, current discussions of digital technologies divide into those that concentrate on the splendid innovations that have been or will be introduced, and those that remind us that they may raise ethical issues. The Centre for Data Ethics and Innovation recently announced by DCMS is clearly focused on innovation and future commercial promise, although ethics gets – well – a nod. This is not unusual. All too often ethical issues are mentioned, but we are offered only bien pensant comments. I am all for better digital education and better online conduct and less disagreeable online stuff: but gestures are not enough and realistic thinking about the ethical and legal questions and about the benefits and the harms that uses of digital technologies bring are needed, and these may require difficult legal, regulatory and deeper cultural changes, and some  may prove irremediable.

The Ada Lovelace Institute will matter because it aims to go beyond reassurance that the ethical issues matter to addressing them. This will not be easy, but it is necessary. I am encouraged when I remember that many of the topics with which the Nuffield Council on Bioethics engaged have been difficult, and that some of the difficulties have been removed or mitigated by penetrating and realistic analysis, and by subsequent action. So I shall comment on a few issues with which I suspect the Ada Lovelace Institute, working with others, including the Turing Institute and the Centre for Data Ethics and Innovation, will need to grapple.

Topics and Questions

Act or Content?

Ethical norms or standards bear on action, and those that are relevant to what is said or done using symbols bear on  speech acts, rather than directly on the semantic content of those acts. Ada Lovelace pointed to questions about  how to do things with words and other symbols, about how to use and manage data —the title of the Royal Society and British Academy report from 2017 gets it right by referring to data management and use. (For that reason I think that the portmanteau phrase data ethics should be used with caution).

History

We may be able to learn something from historical parallels. Plato reports that Socrates thought writing (then a newish technology, which many could not decipher) was a deceptive and defective way of communicating. Unlike the spoken word, written words can be separated from their authors – the speech content can be separated from the speech act – and that separation may mean that nobody stands ready to interpret, explicate, defend or vouch for their meaning, their truth or their trustworthiness. Yet writing later earned its spurs as an ethically reputable technology that allows a permanent trace, indeed as ideal for establishing the provenance and authorship, for keeping reliable records and many other purposes.

But history also provides examples of how difficult it is to deal with new technologies. The invention of printing led to massive disruptions, and it took a more than two centuries to create the legal and cultural structures that defined the roles of printers and publishers, and clarified responsibilities and liabilities. We also know that between the wars the emergence of broadcasting led to massive disputes about its regulation and control. So we should not be surprised that transitions to digital technologies too are leading to massive disputes.

We need to remember that some supposed remedies do not work. The ethics of speech acts cannot be done by trying to distinguish permitted from forbidden content, although this has often been tried. Making words taboo and censoring content simply does not work. We need to focus on what people can and cannot, should and should not, do with data, rather than on the data.

Normative reasoning in trouble

However, we may not be thinking broadly enough about the normative standards – both epistemic and ethical standards – that matter for communication. Since Universal Declaration of Human Rights and the European Convention of Human Rights there has been huge emphasis on rights and neglect of those duties that lack counterpart rights, and specifically on two liberty rights that matter for speech: rights to freedom of expression and to privacy. Other traditional standards that matter for speech – accuracy and honesty, clarity and respect for evidence, civility and decency – are too often ignored or taken for granted, or seen as relevant only for certain sorts of speech act.

As it seems to me, this narrowing of ethical concern, which focuses on recipients rather than agents, forms part of wholesale reversal of ethical perspective in the twentieth century. It is apparent in a lot of attention to liberty rights, and a lot of silence about traditional duties that lack counterpart rights. The Ada Lovelace Institute will – I think – need to consider the full range of normative issues, and not just those that were picked up in the human rights instruments.

Beyond communication

Indeed it may be necessary to think about many sorts of action other than communication. We often speak as if digital technologies were used only to communicate, indeed to call them  communication technologies. Yet they are not used only to communicate, let alone to communicate accurately or responsibly, but also to influence others by targeting them with certain sorts of content and shielding them from other sorts of content. A few years ago it was widely believed that we were on the brink of a widening of communication and participation that would spread democracy and opportunities for participation. The libertarian outlook of many of the leaders and owners of big tech companies was forged in that world. Now, I think the age of the cyber romantics is over, and we need to consider soberly what can be done to ensure that these technologies support rather than damage “a speedy and accurate practical application for the purposes of mankind”.

Anonymisation

Finally I think that Ada will need to think hard about anonymisation. Anonymisation is important for certain very specific circumstances, such as investigative journalism under oppressive regimes.  But it also enables hidden persuasion and manipulation, not to mention cyber-bullying and defamation, and undermining of cultural standards and democratic politics. Digital technologies permit anonymisation and indeed impersonation on a vast scale, and their use is not always benign.

Users of digital technologies have yet to earn their ethical spurs, and it will be and should be hard to earn them. My hope is that the Ada Lovelace Institute will help them to do so.

  • This speech was first delivered at an Ada Lovelace Institute event held at the Nuffield Foundation on 4 December 2018.
  • Read Dr Stephen Cave’s speech from the event >