I do not expect robots to arrive at our door-step and register a new-born baby anytime soon, but it could be fascinating to explore how AI could augment natural human effort in this area. In fact, efforts are already underway.
National legal frameworks and policies around birth registration exist in every country today, but practices remain localized and often manual. In this context, AI is already being tested to extract key information from hand-written documents, often from legacy records stretching back to decades, and create digital archives, that are more easily accessible and amenable to detecting errors, inconsistencies and possible fraud. Using mobile technology, AI is also being put into the hands of field personnel and bringing them closer to where births are taking place, thus facilitating timely initiation of the birth registration process.
Such pilot initiatives are currently being led by inter alia UNICEF and UNFPA. Even AI-powered smart-bots, supporting health workers in the field for maternal and newborn health care, besides birth registration itself, are already under trial.
But what is it that AI can contribute best to birth registration?
AI, as of now, seems overwhelmed with misplaced expectations. It has also preoccupied many of us with apprehensions whether AI has the human touch or not, a question that will ultimately be answered sometime in the future, as much by metaphysics (study of being) as by semantics (language). However, the power of AI as we know it today, rests in epistemology, the study of knowledge.
To digress a bit, epistemology and AI come together in many ways, on how knowledge is acquired, represented and utilized, and how humans engage and consume AI generated information. AI systems are designed to process information, learn from data, and make inferences, essentially acting as tools for acquiring and manipulating knowledge. This positions AI as an epistemic technology, influencing how we understand and interact with the world through knowledge.
However, all this comes with caveats such as: have we the most appropriate model for knowledge representation, and how reliable, trustworthy, transparent, and understandable is the AI generated information?
There are also deeper concerns. How does the use of AI affect our reliance on each other for knowledge and the social dynamics of knowledge creation and dissemination? How can AI contribute to our understanding of knowledge itself, potentially offering new perspectives and insights?
Lastly, AI systems are often viewed to act as epistemic authorities, influencing our beliefs and knowledge acquisition. The interaction between humans and AI systems raises questions about epistemic responsibility, trust, and the potential for bias and manipulation. AI-generated content therefore constantly raises questions about the nature of truth and the role of human judgment in evaluating information. In essence, AI’s development and deployment raise fundamental questions about what constitutes knowledge, how we acquire it, and how we evaluate its reliability. All this makes the study of AI deeply intertwined with the field of epistemology, and its current use must therefore be guided by such concerns[1].
In this light, how can AI support and improve birth registration practices?
Birth is a vital civil registration event for providing legal recognition of a child’s identity, nationality, and age. It is also critical for establishing citizenship and exercising the associated legal rights, as well as essential for accessing education, healthcare, and other state services. In turbulent times it could be critical for determining refugee status or when faced with extreme situations, leading to statelessness for example. Furthermore, as is well-known, civil registration data when aggregated into Vital Statistics (together – CRVS) is one of the pillars of effective governance, planning public services, and addressing social challenges, through systematic assessment of evolving population characteristics.
Among other AI applications involving birth registration, especially two could be of decisive importance. Besides being a vital event, a birth is also a node in a network of relationships, of which a genealogical chart, or the family tree, is one representation. Birth registration also affixes parentage for example, an important piece of information required for citizenship determination, as well as extreme situations such as family reunification.
It would especially be handy in other unusual situations, like those involving naturalization and adoption, as well as more modern means like surrogacy and in vitro fertilization. Such births then need to be mapped onto legal frameworks that could even change with time. An existing statute might grant citizenship to a new-born whose both parents are citizens. This could be amended, for example, to all or some grand-parents being citizens as well.
National ID programs, as introduced by numerous countries, with its unitary definition of each individual, excludes explicit definition of relationships, which then require ad hoc legal procedures especially where inheritance or succession needs to be determined. Neither national ID programs, nor birth registration cover this comprehensively and often requires an intelligent and diligent search through established familial networks that could be extremely time consuming. However carefully devised AI mechanisms could effectively and efficiently cover this area.
However, the entire gamut of issues that birth registration needs to address, legally and socially, often extends to more complex legal definitions and positions. Many countries for example, take variable positions vis-à-vis “legal” parent and a “biological” parent. Similarly, country of birth and countries of parents’ nationalities could also affect where an individual stands on various counts, such as eligibility of certain services and welfare as well as nationality, obtaining a passport etc. Many such cases go into litigation needing reference to past documentation as well as searching for precedents. However, this is well within the possibility of AI and if deployed rapidly and reliably, could bring relief to a large number of individuals that have to often wait for years for their status to be determined. This principle could equally well be applied to refugee status determination, for example.
For this to be become a reality via the use of AI, would need two critical ingredients. One, that all the required information is available reliably somewhere, where AI can find and organize it, and second, the legal frameworks are conducive for this to happen. For example, GDPR could prevent such searches to be made as they violate individual privacy, as understood today. However, as shown in the popular Swedish TV serial, The Breakthrough, such information processing, coupled with DNA samples could be the last resort in solving a murder case, even 16 years after.
Birth certificates also serve as breeder documents for other ID documents, such as passports and drivers licenses. How does AI strengthen and reinforce all this, also needs to be considered.
It must also be noted that the advent of AI comes simultaneously with digital technology being considered for birth registration. However, as the primary definition goes, it must result in a record that is universal, continuous, compulsory and permanent. Today often diverging views are being held whether digital records are permanent or not. It would seem that the jury needs to sit and make similar assessments about AI as well.
Yet another aspect of birth registration is predicting how many and even where births will next occur. Today, it is possible to make approximations that could well be improved using AI. All this helps in mobilizing public health services. Also, despite improved practices, differences between CRVS and census counts persist that need to be accounted for, that is yet another area that AI could effectively address by complementing existing statistical sampling methods.
As such, these are only some examples, and over the years, many more uses could be found for deploying AI constructively around birth registration. However an important aspect here is that such effort should be beneficial and not harmful in any way. Therefore, as always, sufficient safeguards should be put in place, legally, procedurally and technically to prevent any harmful use.
As can be seen, these various potential uses, span a wide range of AI tools as we know them today, such as machine learning models (with or without supervised learning as well as reinforced learning), deep learning models (neural networks), generative models (large language models like GPT) as well as others that still wait to be discovered.
Would AI one day predict who would be born, to whom and where? Today this is only the subject to mythology and story-telling. But tomorrow, who knows? That then is also the difference between the epistemological and the metaphysical of AI. In today’s exciting times of rapid AI developments, its application to birth registration could indeed prove to be beneficial after all, but one must also be watchful of any harmful and evil intent.
[1] All this and much more is currently being discussed and contemplated by scholars, as in the recent work on the philosophy and ethics of AI (Alvarado, AI as an Epistemic Technology). Understanding AI as an epistemic technology also has significant implications for important debates regarding our relationship to AI technologies. This paper provides an overview of such implications, particularly those pertaining to explainability, opacity, trust and even epistemic harms related to AI technologies.
Sanjay Dharwadker 
Leave a comment