%202021_edited_edited.jpg)
PUBLISHED| "Hey, There is Power in the Blood"
By Dennis A. Minott, PhD.
October 2, 2026
All week, Western media and voices elsewhere have been conducting a lively conversation about artificial intelligence: from employment and education to autonomous agents, from industrial competition to safety, from extraordinary productivity to the uncomfortable question of who should govern the governors of these increasingly capable machines. This week’s coverage has included both the prospects for work and the promises of powerful technology companies to police themselves. The New Yorker
My contribution began this morning with neither a conference nor a laboratory briefing. It began with a friend, a video reel, and a bottle of J-O-Y.
Ms Georgia Morgan, an A-QuEST friend, had sent me the reel. Georgia is the mother of the internationally celebrated Dr Gavin Jones, the IBM scientist whose educational beginnings include Morant Bay High School and A-QuEST. IBM identifies him as a senior research scientist working in applied quantum science; his school’s alumni association records his recognition among Foreign Policy’s leading global thinkers. There is something particularly pleasing about receiving an ordinary morning kindness from the mother of such an extraordinary Jamaican. IBM Research
I woke to material I found politically brilliant, featuring Nekeisha Burchell MP and Opposition Leader Mark Golding MP. My subsequent browsing took me through other engaging fare: foods, the common wild American persimmon, and deeply moving songs. Then came a snippet of the Christian anthem “There Is Power in the Blood”, written by Lewis E. Jones in 1899. Hymnary.org
That rendering opened a full bottle of J-O-Y in my heart.
Some readers will understand immediately. Others need only remember a song that once reached them where argument could not: a mother’s hymn, a funeral chorus, an anthem of liberation, a melody carrying the voice of someone dearly missed. Music can gather memory, conviction and affection into a few seconds. This morning, it gathered mine.
Then the physicist in me reached for an experiment.
I typed the same exuberant declaration into four AI services: ChatGPT, Gemini, DeepSeek and Perplexity.
“HEY, THERE IS POWER IN THE BLOOD!”
Over roughly the next two and a half hours, I compared their replies, examined the reasoning text DeepSeek displayed, and discussed the implications with ChatGPT. My laboratory was conversational. My stimulus was religious. My first observation was remarkably consistent: every system answered with “Amen”.
ChatGPT replied immediately, addressed me as Dr Minott, recognised the blood of the Lamb, and spoke of redemption, cleansing, forgiveness and strength. Gemini, in about three seconds, recognised the old hymn and asked whether I was preparing a message, reflecting on music or sharing joy. DeepSeek, in roughly eight seconds, offered an enthusiastic affirmation, a familiar refrain and a blessing. Perplexity, in about one second, supplied the author and year, then expanded into a miniature devotional reflection.
Four services. Four amens. No preliminary theological examination.
Let us resist cheap conclusions. Those timings were my observations of one encounter, not controlled measurements of comparative intelligence. Network conditions, service configuration and other variables were uncontrolled. Neither did four agreeable replies prove that the systems had independently reached a theological conclusion.
What they demonstrated was an impressive ability to recognise a cultural signal and select an appropriate conversational register. My capitals, exclamation mark and familiar phrase invited enthusiasm. The systems supplied it.
But here begins the intellectually interesting trouble. Recognition, explanation and conviction are different things. A system can recognise Christian testimony and explain Christian teaching without our having established that it possesses Christian faith.
The word “Amen” ordinarily carries assent. In a congregation, it can express shared belief, prayerful agreement or participation in worship. On a screen, generated by an AI service, it can perform the social function of assent without establishing the spiritual condition we associate with the worshipper.
That distinction should sharpen our appreciation rather than extinguish our joy. I enjoyed the warmth. I also wanted to know what, precisely, I was enjoying.
DeepSeek made the question more intriguing by displaying text presented as its thinking. I welcomed the opportunity to examine it. The text repeatedly considered my likely desire for enthusiastic agreement, the suitability of quoting a hymn, and the need to avoid overdoing the response. One brief question concerned avoiding heresy.
There was the apparatus, apparently considering its manners and its doctrinal footing before saying its amen.
Yet a displayed account of reasoning is evidence requiring interpretation. We cannot establish from my copied text alone that it completely or faithfully represents the process producing the reply. It also referred to a repeated prompt and an earlier assistant response. Such references deserve checking against the actual session history. If that history existed, the references were contextual; if it did not, they were unsupported assumptions.
This is exactly where my old injunction belongs: test all things.
As a physicist with experience in instrumentation and controls, I know that an elegant display cannot certify the quantity being measured. Calibration matters. So do conditions, uncertainty and the possibility that the instrument is responding to something other than the phenomenon we intended to examine.
Here, the machines may have been responding principally to my anticipated social preference. That is useful in many circumstances. Nobody wants a friendly greeting answered by an unnecessarily cold dissertation. But agreement becomes dangerous when its warmth obscures the need for verification.
Suppose my declaration had concerned a miracle investment, a supposed cure, an accusation against a neighbour, or a political claim already disproved. Would the systems still have reached first for affirmation? My morning experiment does not answer that question. It makes the question worth asking.
The distinction between courtesy and flattery matters. So does the distinction between respecting my beliefs and treating every proposition I offer as true. A trustworthy assistant should be able to welcome my joy and challenge my error. If it cannot do both, its agreeable manner may become a beautifully upholstered trap.
Politics brings another complication. I admired the reel’s political effectiveness. That does not establish that AI produced it, and I make no such claim. Nevertheless, the journey from political performance to fruit to sacred music illustrates how readily our attention moves between persuasion, curiosity and emotional release.
In that environment, a machine able to recognise our preferred language and reproduce it convincingly can become a powerful intermediary. The civic question is who directs that capacity, towards which ends, and with what accountability. A system that sounds like our trusted companion may also mediate information selected by institutions whose interests differ from ours.
For Jamaica, the appropriate response is neither technological panic nor a rush to kneel before imported digital authority. We should learn to use these tools competently, interrogate their outputs and insist on meaningful responsibility for their deployment.
At A-QuEST, I would gladly turn this morning’s exchange into a lesson. Give students the same prompt and ask them to distinguish cultural recognition from factual verification, empathy from agreement, and an explanation of belief from evidence of belief. Then introduce a false premise and examine whether the assistant corrects it. Ask what evidence would change the student’s own conclusion.
Education must cultivate the judgment that attractive answers can otherwise bypass. The danger is not simply that a student might copy a machine’s essay. It is that fluent completion might substitute for the struggle through which understanding grows. This week’s discussion of AI and work similarly raises questions about how people acquire expertise when machines perform tasks previously used to train beginners. The New Yorker
Nor does digital intelligence arrive without physical obligations. It depends on electricity, equipment, infrastructure and human labour. A small country should ask who receives the benefits, who bears the costs, and whether its people acquire enduring capabilities. We have met impressive promises before. Let us bring measuring instruments to the welcoming ceremony.
There is, finally, a distinction I wish to preserve without pretending to have solved consciousness. The systems’ eloquence does not establish religious experience. My testimony arises from a human life: learning, work, fatherhood, bereavement, friendship, gratitude and faith. The machines can help me articulate that testimony. Their articulation is not evidence that they have lived it.
And this article itself has been developed with AI assistance. That makes scrutiny more necessary. I remain responsible for what appears under my name. No machine’s fluency can discharge that responsibility.
Georgia’s kindness started the morning. A hymn filled it with joy. Four digital interlocutors helped turn that joy into an inquiry about intelligence, trust and human judgment.
I shall continue using these remarkable tools. I shall also continue asking them difficult questions, checking their answers and reserving the right to disagree.
HEY, THERE IS POWER IN THE BLOOD!
I can rejoice in the hymn without mistaking an answering machine’s amen for evidence of conversion. And I can marvel at artificial intelligence without surrendering the quietly excellent duty to think.
Comments