Read the Commentary How to Use AI Responsibly

As I was growing up, we valued our freedom- of movement and use. Taking risks was just a part of life.  We didn’t have car seats. I would have laughed if you had asked me to wear a helmet when riding my bicycle.  Each technological change comes with both risks and rewards. AI has both secular and spiritual risks.  The greatest of these you may not immediately recognize. It is spiritual.  For example, if I had AI write this Commentary for me, it would be morally deceitful—( it might do a better job, but it wouldn’t be me).

Suppose you just ask your AI app a question—using the search part of your browser—instead of using your own memory and experience . How do you know the answer is true and trustworthy? How do you know the large language model that gives you an answer has any deep understanding of a problem? You don’t unless you verify sources.  AI should be an aid not the ultimate authority.   Use can weaken memory, reasoning, and discernment if not careful.

 I turn now from AI that answers questions to AI that helps carry out a process. Suppose you use  AI as a multi-step process for deep research, or as an assistance agent. An assistance agent example would be comparison shopping with recommendations.  You just outsourced your personal judgment to the machine if not careful.   To mitigate risk, carefully bound what you are asking for, give the least permissions, and retain frequent human checkpoints.

AI helping with  secular judgment is a different process to an  AI participating in action and transactions. If you use your browser as a transactional agent ( e.g.) to buy something the risk is greater.  Beside obvious security risk, such convenience can cause you to lose sight of the fact that many transactions have significant moral dimensions.  Thus, prudence and the effects on the conscience are eroded.  To mitigate risks requires approval before each transaction and avoids using  AI when wrong results could be disastrous.  

You are unlikely in the near term to use autonomous agents or multi-agent systems directly.   Autonomous agents replace human stewardship. .  Multi-agent systems make it hard to determine accountability and can be heavily error prone.  In particular they can take on God-like characteristics with you as the worshiper. They can make you slave-like with the agent as the master.  To mitigate risks,  reduce the scope of any action, make the agent maintain detail logs, and forbid any independent high stakes action In any case, a human must be in control with clear audits of transactions.

I turn now from normal technical and practical dangers to the deepest moral and spiritual issues we face. Where are these greatest spiritual concerns?  Trusting and AI system to give  you guidance on any moral question is easy to do and hazardous.  AI large language systems do not contain the wisdom found in the Word of God. Take an easy moral question: Is homosexual activity permissible for a human being?  Here is a direct quote of one answer when I asked this question of ChatGPT: “From an affirming Christian standpoint, it can be regarded as permissible, because those Christians argue the biblical prohibitions address specific exploitative or idolatrous situations rather than faithful same-sex relationships as understood today.”  So, you can find any position you may wish to take.

This doesn’t mean one cannot responsibly use AI to search biblical resources, biblical scholars, or even scholarly commentaries and interpretations.  But you must know the source and not “out-source” your judgements and biblical wisdom for a source that is non-spiritual and non-human.  Any questions of conscience,  spiritual truths, doctrine, Christian practices, judgement and trust in God must never be off-loaded to an AI agent.

My secular guidance for using AI would be simple: always verify its sources, give it only narrow permissions, define carefully what you are asking, be especially cautious with medical and financial matters, and keep the human being firmly in control at every stage. Spiritually, the boundary should be even clearer. AI may assist with research, organization, and drafting, but it should never be allowed to make judgments on matters of faith, conscience, doctrine, or spiritual counsel.