Wednesday, October 25, 2017

Ethical tech adoption in civil society

Most digital technologies are designed by, or at least brought to mass adoption by, commercial enterprises. This is often, but not always done, on the back of government funded infrastructure or research. Civil society, which exists as a counterbalance to and with some intended separation from, both markets and governments, often adopts new technologies without first considering how the tools might conflict with the sector’s own values.

Some technological approaches, such as artificial intelligence (AI), have attracted enough public detractors that industry is responding with its own policy association “principles of practice,” bringing an undeniable stamp of “regulation pre-emption.” AI, of course, has been in the public’s mind for decades, courtesy of robot cartoons and science fiction. There's a robust debate among tech leaders about the future of AI.

For other technologies, where the need for placating the public is less urgent, the typical deployment strategy goes something like this:
·      R&D, either in universities or industry
·      experimentation in specific applications,
·      commercial take up where experimentation is successful
·      rampant application to broader business opportunities,
·      crossover experimentation on social issues
·      belated social sector response when application generates “unintended” (though not necessarily unpredictable) consequences

Is there a better way to do this? Can the social sector pre-emptively develop a set of guardrails for the application of new technologies so that predictable harm (at least) can be minimized or prevented? 
Doing so requires articulating a set of sector values that would apply to multiple technologies, or at least a means of checking technologies for fit not at the “shiny object” level but at the mission-alignment, core values level. There are some such efforts to do so - at least one in AI and public services and the responsible data principles could be seen as a digital data level version of this.

Consider the blockchain. The technology's protocols were originally developed as a means of enabling trust and accountability in a decentralized manner. The first application to gain popular attention were currencies and currency exchanges. Now, the blockchain is being used (or proposed to be used) for other types of trusted exchanges that require some form of independent accountability.

In order to function without a central repository, the blockchain requires the creation of a permanent record of an action which is verifiable by the larger network.

Those technological requirements result in a few features that have come to dominate public discussion of the blockchain. These include, but are not be limited to:
·      It is immutable. Once a piece of information is added to the chain it cannot be changed.
·      It is decentralized and verification is built into the technology. There is no single point of control.

It is these technological features that need to be assessed against the values of purpose of a particular task or action. Is immutability of record a good thing? Is it in line with the goal seeking to be achieved? If the action being taken involves tracking material goods in a supply chain than the the answer may be yes. If the action being taken involves tracking a human being through space and time, then the answer is not as straightforward. It’s easy to imagine cases where a person might not benefit from a permanent record of their presence – escaping violence, seeking assistance to which stigma is attached, peaceably protesting injustice to name a few.

Now let’s consider the other commonly pitched feature of blockchain - decentralized verification. If there is no single point of control for governing the system, then there is also no point of redress for an individual who may be wronged by it. Since “social good” often centers around changing dynamics between individuals and systems (think education, health care, disaster relief, migration rights for just a few examples), applying a system that provides no redress for individuals is unlikely to be seen as an improvement (at least by those individuals supposedly being helped).

Social sector applications of new technologies need to consider the tradeoffs in values between the mission being pursued and theencoded values of the technology itself. Business applications of new technologies are often focused on the commercial prerogatives of efficiency, scale, or cost, and the primary perspective is that of the implementing organization. Social good applications must align with a significantly more diverse, complex, and structural set of values, while not compromising the rights of the people theoretically being assisted.

Civil society needs to adopt and adapt to the digital age we live in. Many technological applications are appropriate. But in assessing these opportunities, we must consider not just each new and shiny technology but also the values they encode. And the social sector should assess this alignment in relationship to the rights and opportunities of the intended beneficiaries, not to the organization implementing the technology.

No comments: