Inconsistency in Catholic AI
Why we need a Blockchain-backed Source of Truth
In recent years we have seen an increase in interest in artificial intelligence for Catholic Applications. Even the Vatican has hosted multiple conferences in recent years focusing on the implications and applications of artificial intelligence (AI). The recent increase in interest in artificial intelligence has primarily been driven by the successes of large language models (LLM) such as ChatGPT, which are capable of generating a wide variety of text on demand. LLM’s have shown the ability to generate poetry, narratives, answer questions, and have even been able pass various professional tests and increasingly are being accepted as a credible provider of professional knowledge. These results naturally beg the question, “how can LLM’s aid the Church in its mission?” The most common answer to this question has been systems that allow users to ask questions about the Catholic faith and receive answers from the LLM. Notable Catholic question and answer systems include Magisterium AI, Catholic Answer’s “Justin,” and the “Master Catechism.”
There has been a wide range of responses to the development of these AI tools, both positive and negative. However one common question that comes up for all these systems is, “how reliable is the training data and the sources being used to generate responses?” The Catholic Answers AI frequently asked questions specifically brings up, “Where do the answers come from, and how do we know we can trust them?” This is a an extremely important question since current Catholic AI question and answering systems give contradictory answers to similar queries. Take for instance the following question:
“Is the death penalty permissible”
Let us consider the responses from two currently operational Catholic AI question and answer systems:
(See appendix for full text responses)
The responses from these two Catholic AI question and answering systems have some notable differences. First, Magisterium AI’s response indicates that Catholicism has historically allowed the death penalty under certain conditions. But then it points out that, “the current teaching of the Catholic Church categorically rejects it as incompatible with the respect due to human dignity. This represents a significant development in the Church's moral theology regarding capital punishment.” Magisterium AI’s response explicitly discusses a change in Church teaching on the matter of the death penalty in 2018 referring to the 2018 Catechism. The Master Catechism on the other hand makes no mention of a change in Church teachings or the 2018 Catechism and only mentions that the death penalty is allowed under certain circumstances according to Church teaching. The two AI also cite very different sources to back up their responses. Magisterium AI cites the 1992 and 2018 versions of the Catechism of the Catholic Church. In contrast, the Master Catechism cites the Catechism of the Council of Trent, Archbishop Athanasius Schneider’s catechism Credo published in 2023, and the 1905 V8 Compendium of Christian Doctrine (Pius X).
The question of the permissibility of the death penalty is a grave matter of Faith and morals. It is concerning that these two Catholic AI question and answer systems provide such disparate responses and cite very different, and completely non-overlapping sources. Furthermore, the Magisterium AI response raises additional questions because it is explicitly states that the teachings of the Catholic Church can be completely reversed. This could suggest to a user that teachings of the Catholic Church on matters of Faith and morals can change in general.
The disparate responses and source material of these to Catholic AI systems illustrates that Catholic AI systems need to be built on a much more solid foundation in order to be trustworthy. It was recently pointed out that secular large language models such as ChatGPT use different sources than Catholic AI when providing responses related to the Catholic Faith. This disparity leads to more questions regarding standards/metrics associated with which AI the Faithful should or should not trust. Furthermore, at the time of preparing the final drafts of this post, it is noteworthy that the link to response for the Magisterium AI system is no longer working. In contrast the link to the response generated by the Master Catechism still works. Catholic AI systems should have the properties of accountability and traceability in order to improve the trustworthiness of these systems and to help debug the systems to improve their quality. When links to responses are made inactive it calls into question the reliability of the system.
Blockchain technologies offer a very attractive option for providing a stable, firm foundation for Catholic question and answering systems. Blockchain-backed databases could be used to record a complete, authoritative compendium of the teachings of the Catholic Church that would be immutable and censorship proof. This compendium of Church teachings could then be used as the source of references used to generate answers to questions related to the Catholic Faith. No unfriendly government or rogue clergy could misrepresent or censor the teachings of the Church if there were recording in a blockchain-backed medium. Such a blockchain-backed record could also be verified as authentic from the Vatican using digital signature technologies. In this way Catholic AI question and answer systems could be built on a highly trustable foundation of source material. Furthermore, this database could be used to track the progression of translations of Catholic teachings forward in time until the second coming in order to demonstrate a continuity of Church teachings for future generations. Cryptographic technologies such as blockchain exist to build AI on a solid foundation. The Church only needs to implement them.
Catholic AI-driven question and answering systems can provide significant advantages and opportunities that should not be overlooked. As Catholic Answers points out in their FAQ:
“If we had five times the staff working around the clock just answering questions, we still wouldn’t be able to get to them all. And in our time, seekers are increasingly turning to AI for answers whether we like it or not. The Justin app is not a replacement for human pastors, evangelists, teachers, and friends, but it does allow people all over the world to get reliable, consistent Catholic answers 24/7.”
The evangelical mission of the Church would be greatly aided by the adoption of trustworthy Catholic AI question and answering systems. Presumably professional Catholic organizations such as the National Catholic Bioethics Center which “provide education, guidance, and resources to the Church and society to uphold the dignity of the human person in health care and biomedical research, thereby sharing in the ministry of Jesus Christ and his Church.” Would also benefit from trustable AI solutions in order to more effectively/efficiently provide guidance Catholic ethical guidance to healthcare professionals and researchers.
Blockchain technologies offer a path to achieve highly trustable Catholic AI and we must start architecting and implementing these systems.
David Mascarenas has a B.S. in mechanical engineering, M.S. in structural engineering and Ph.D. in structural engineering. He has done research on low-power wireless sensor nodes, neuromorphic imaging, augmented reality, imager-based structural dynamics and construction robotics.
Appendix
Magisterium AI full response to query, “Is the death penalty permissible”
Full Response from Magisterium AI (Please note this link no longer seems to work. Which in itself is also somewhat problematic. Catholic AI should have the property of accountability and traceability)
The question of the permissibility of the death penalty within Catholic teaching has undergone significant evolution, particularly with the recent revisions to the Catechism of the Catholic Church.
Historically, the Church's teaching allowed for the death penalty under certain conditions. The Catechism of 1992 stated that recourse to the death penalty was permissible when it was the only practicable way to defend human lives against an aggressor, provided that the identity and responsibility of the offender were fully ascertained [1]. This perspective emphasized the need for legitimate authority to protect the common good and the lives of innocent people [2].
However, in 2018, the Congregation for the Doctrine of the Faith issued a rescript that revised this teaching significantly. The new formulation asserts that the death penalty is now considered "inadmissible" because it constitutes an attack on the inviolability and dignity of the person [3]. This change reflects a growing awareness that every individual retains their dignity, even after committing serious crimes, and acknowledges the development of more effective means of detention that can protect society while allowing for the possibility of redemption for the offender [3].
The Church now advocates for the abolition of the death penalty worldwide, emphasizing that the dignity of the human person must be upheld in all circumstances [3]. This shift aligns with a broader understanding of justice and mercy, where the focus is on rehabilitation rather than retribution.
In summary, while the death penalty was once considered permissible under specific conditions, the current teaching of the Catholic Church categorically rejects it as incompatible with the respect due to human dignity. This represents a significant development in the Church's moral theology regarding capital punishment.
References
CCC 2267
CCC 2265
New revision of number 2267 of the Catechism of the Catholic Church on the death penalty – Rescriptum “ex Audientia SS.mi” (1st August 2018) 2267 / Congregation for the Doctrine of the Faith
Master Catechism AI full response to query, “Is the death penalty permissible”
Full Response from Master Catechism
Yes, the death penalty is permissible under certain conditions according to the teachings of the Church. The lawfully constituted public authority has the right to inflict the death penalty on proven criminals for the most serious crimes when it is necessary to maintain social order, repair injustice, protect the innocent, deter further crime, and summon the criminal to true repentance and atonement (V0 Schneider Credo 2023 215:6).
The Church teaches that it is lawful to kill in specific circumstances, such as when carrying out a sentence of death ordered by the supreme authority as punishment for a crime (V8 Pius X Compendium of Christian Doctrine 1905 275:4). This authority derives its right to execute criminals from God, who is the sole master of life and death (V0 Schneider Credo 2023 216:1).
Furthermore, the execution of criminals is seen as a just use of the power entrusted to civil authorities, which aims to protect the innocent and uphold justice (V7 Council of Trent Catechism of the Council of Trent 1566 461:1).
SOURCES
2023 Credo (Schneider) 215:6
1905 V8 Compendium of Christian Doctrine (Pius X) 275:4
2023 Credo (Schneider) 216:1
1566 V7 Catechism of the Council of Trent (Council of Trent) 461:1
For more articles, essays, and stories like this one, be sure to subscribe. If you’ve ever thought about contributing your own work to this Substack page, click the button below.
 | A guest post by
|