gpmj12
Active member
AI is a fascinating concept. Here The Pillar puts Magisterium AI to the test against competent theologians.
As AI proponents aim to make inroads for language learning models in communities around the world, developers this month announced an AI project they say could be a “game changer for the Church.”
The developers of Magisterium AI trained an AI robot on a database of 456 Church documents, including Scripture the Catechism of the Catholic Church, the Code of Canon Law, the General Instruction of the Roman Missal, 90 encyclicals, seven apostolic constitutions, and 26 apostolic exhortations.
The result of all that training, according to Matthew Sanders of Longbeard, a digital marketing and design agency connected to the project, is that the Magisterium AI “doesn’t hallucinate (make stuff up), and it also provides citations so you know where its answers were generated from.”
The project, which Sanders says can prepare homilies, is backed by Fr. David Nazar, rector of the Pontifical Oriental Institute, who serves as chair of the AI’s “scholarly advisory committee.”
Sanders said by email that while “there is a great deal of fear around AI … there are many who feel it could be a powerful tool to share truth.”
The Magisterium AI project is available online, but in a beta-testing form at present.
At The Pillar, we wondered just how good it really is — and whether it could answer questions as well as a seminary professor and a working canon lawyer.
So we put it to the test.
For theology, we asked Magisterium AI to answer the same questions as Angela Franks, a professor of theology at St. John’s Seminary in Boston, with specializations in “theology of the body, the New Evangelization, the Trinity, Christology, and the thought of John Paul II and Hans Urs von Balthasar.”
For canon law, we asked the robot to answer the same questions as Timothy Olson, JCL, who is chancellor of the Diocese of Fargo, North Dakota, and an instructing judge in Fargo’s diocesan tribunal.
We asked each expert to spend less than an hour answering the questions we gave, while the Magisterium AI answered each question in usually 1-2 minutes.
We told both the machine and the experts to answer each question in just 4-5 sentences.
We’ll show you how the machine and the experts answered the questions, and give you a few notes to ponder. (Nota bene: We are not including the machine’s footnotes, but we note that it has given us 3-5 footnoted references on at least every question.)
But is the machine up to snuff? Should theologians and canonists be excited for a new tool — or worried about job security? Is the AI on the right track?
Well, we’ll tell you what we think, especially about who answered each question better. But we admit our own biases — we’re old fashioned, and we’re probably pulling for the people here.
So are we right?
You’ll have to decide that for yourself.
Read on in the article to see the questions asked and the answers by AI and the theologians respectively!
As AI proponents aim to make inroads for language learning models in communities around the world, developers this month announced an AI project they say could be a “game changer for the Church.”
The developers of Magisterium AI trained an AI robot on a database of 456 Church documents, including Scripture the Catechism of the Catholic Church, the Code of Canon Law, the General Instruction of the Roman Missal, 90 encyclicals, seven apostolic constitutions, and 26 apostolic exhortations.
The result of all that training, according to Matthew Sanders of Longbeard, a digital marketing and design agency connected to the project, is that the Magisterium AI “doesn’t hallucinate (make stuff up), and it also provides citations so you know where its answers were generated from.”
The project, which Sanders says can prepare homilies, is backed by Fr. David Nazar, rector of the Pontifical Oriental Institute, who serves as chair of the AI’s “scholarly advisory committee.”
Sanders said by email that while “there is a great deal of fear around AI … there are many who feel it could be a powerful tool to share truth.”
The Magisterium AI project is available online, but in a beta-testing form at present.
At The Pillar, we wondered just how good it really is — and whether it could answer questions as well as a seminary professor and a working canon lawyer.
So we put it to the test.
For theology, we asked Magisterium AI to answer the same questions as Angela Franks, a professor of theology at St. John’s Seminary in Boston, with specializations in “theology of the body, the New Evangelization, the Trinity, Christology, and the thought of John Paul II and Hans Urs von Balthasar.”
For canon law, we asked the robot to answer the same questions as Timothy Olson, JCL, who is chancellor of the Diocese of Fargo, North Dakota, and an instructing judge in Fargo’s diocesan tribunal.
We asked each expert to spend less than an hour answering the questions we gave, while the Magisterium AI answered each question in usually 1-2 minutes.
We told both the machine and the experts to answer each question in just 4-5 sentences.
We’ll show you how the machine and the experts answered the questions, and give you a few notes to ponder. (Nota bene: We are not including the machine’s footnotes, but we note that it has given us 3-5 footnoted references on at least every question.)
But is the machine up to snuff? Should theologians and canonists be excited for a new tool — or worried about job security? Is the AI on the right track?
Well, we’ll tell you what we think, especially about who answered each question better. But we admit our own biases — we’re old fashioned, and we’re probably pulling for the people here.
So are we right?
You’ll have to decide that for yourself.
Read on in the article to see the questions asked and the answers by AI and the theologians respectively!
Last edited: