Karpathy’s Autoresearch and the Parable of the Talents: What AI Stewardship Looks Like in Practice

A few weeks ago, Andrej Karpathy — former AI director at Tesla, co-founder of OpenAI — released a project that made me think about ministry.

I didn’t expect that either.

Karpathy built a framework called autoresearch. It runs autonomous ML experiments on a single GPU while the researcher sleeps. The AI agent modifies training code, runs a 5-minute experiment, evaluates the result, keeps improvements, discards failures, and loops. About 12 experiments per hour. Roughly 100 overnight. He woke up to measurable performance gains — with zero human intervention during the run.

The part that got me: Karpathy doesn’t write the training code anymore. He writes a Markdown file — plain English instructions — that tells the AI what to research, what constraints to follow, and when to stop. His words: “you are programming the `program.md` Markdown files that provide context to the AI agents.” He calls this “programming in Markdown.”

The human moved up one level of abstraction. Define the methodology, set the guardrails, let the system execute. Not less involved — involved differently, at the level of direction instead of mechanics.

39,800 GitHub stars in the first two weeks. The tech world noticed.

I think the church should too.

The Parable We Keep Skimming

In Matthew 25:14-30 (ESV), Jesus tells the story of a master who entrusts his servants with talents — significant sums of money — before leaving on a journey. One receives five talents, another two, another one. The first two invest and double their resources. The third buries his in the ground.

When the master returns, the investors are praised: “Well done, good and faithful servant. You have been faithful over a little; I will set you over much” (Matthew 25:21, ESV). The one who buried his talent gets rebuked. Not for losing money — he hadn’t lost anything. He was rebuked for doing nothing with what he’d been given.

We tend to read this as a general principle about using your gifts. It is that. But I think there’s something more pointed here for 2026.

AI is a talent in the Matthew 25 sense. It’s a resource placed in front of this generation, and we have a choice. Invest it toward the mission, or bury it because the risk feels too high.

What This Looks Like at My Desk

I want to be specific, because the abstract conversation about “AI and the church” doesn’t move anyone forward.

I’m Director of Product at HarperCollins Christian Publishing, where I lead Bible Gateway — a platform serving over 75 million monthly visitors engaging with Scripture. Before this role, I led product for SermonCentral, which grew to 14,700+ paying subscribers with access to more than 145,000 sermon manuscripts.

Over the past year, I’ve built a system of 18 AI agents that handle competitive analysis, research synthesis, meeting intelligence, content drafting, and task management. Several run overnight — not unlike Karpathy’s loop. The architecture is different (mine orchestrate across business functions, his optimizes a neural network), but the pattern is identical: define methodology, set constraints, let the system execute, review results in the morning.

Every hour I used to spend pulling competitor data or formatting reports is now an hour I spend thinking about how 75 million people experience Scripture online. Or how to make Bible Gateway better for the person opening it at 2 AM because they can’t sleep and need something solid to hold onto.

Karpathy programs research methodology in Markdown now instead of writing Python. I program strategic priorities and agent instructions instead of pulling spreadsheets. The abstraction layer moved up. The work got more human, not less.

The Fear Is Understandable — and Partly Right

I hear the concerns from church leaders, and I take them seriously.

AI will replace authentic ministry. AI will make pastors lazy. AI will simulate relational presence that only a human body in a room can provide. These aren’t irrational. Some are already happening in small ways.

If a pastor uses AI to generate a sermon they never wrestle with, that’s a problem. If a church deploys a chatbot as a substitute for pastoral counseling, that’s a problem. If we treat AI-generated prayers as equivalent to the honest, stumbling prayers of a person before God — we’ve lost something that matters more than efficiency.

But Karpathy’s work shows the other path. The tool doesn’t replace the human. It moves the human to where they’re most needed.

The pastor doesn’t stop preaching — they stop spending 4 hours hunting for the right illustration and spend that time with the family walking through a divorce. The administrator doesn’t stop managing — they stop updating attendance spreadsheets and spend that time training volunteers. The ministry leader doesn’t stop leading — they stop drowning in email and spend that time on the phone with a donor questioning their faith.

I’ve lived this tradeoff. When my agents took over competitive analysis (something that used to eat 3-4 hours a week), I didn’t fill that time with more busywork. I spent it in 1-on-1s with my team and in deeper product strategy. The output quality went up because I was operating at the right level of abstraction.

Where the Line Is (and Where I’m Still Figuring It Out)

I want to be honest — I don’t think anyone has this mapped perfectly yet. I certainly don’t.

Here’s where I’d draw it today:

AI should handle the administrative. Scheduling, data analysis, report generation, email triage, content formatting. These consume enormous amounts of ministry time, and they don’t require pastoral presence. Automate them aggressively.

AI should accelerate the research. Sermon prep research, theological cross-referencing, community demographic analysis. These benefit from AI’s speed and scope. The pastor still does the synthesis — the “what does this mean for my people on Sunday” work. But raw material gathering? Let the machine run overnight, like Karpathy’s experiments.

AI should never simulate the relational. It should not write your prayers. It should not be the voice your congregation hears when they need a shepherd. It should not replace the hospital visit, the awkward conversation in the parking lot, the moment after the service where someone says what they’ve been carrying for months.

The servant in Matthew 25 who was praised put the resource to work — but in service of the master’s purpose, not his own convenience (Matthew 25:20-23, ESV).

Here’s the tension I haven’t resolved: where does “accelerating research” end and “simulating thinking” begin? When an AI summarizes 30 commentaries on a passage, is the pastor still doing exegesis, or are they just picking from a menu? I don’t have a clean answer. I think it depends on whether the pastor is engaging the summaries critically or just grabbing the first one that sounds good. But that’s a discipline question, not a technology question — and discipline questions are harder to solve with guardrails.

If You’re a Church Leader Starting from Zero

You don’t need 18 agents. You need one tool that saves you 3 hours a week.

Pick the task that eats the most time with the least relational value. For most pastors I’ve talked to, it’s sermon illustration research, email management, or meeting notes. Start there. Learn one tool well. Measure the hours you get back.

Then — and this is the part most people skip — reinvest that time in something only a human can do. A visit. A phone call. An hour of prayer you’ve been meaning to protect but kept losing to administrative drift.

Set your guardrails before you need them. Write down what AI will not do in your ministry context. Revisit it quarterly. Technology expands into unintended spaces when boundaries aren’t explicit — I’ve watched this happen in product development for 15 years.

The Talent in Front of Us

Karpathy’s autoresearch is an engineering achievement. But the deeper pattern is almost theological: the human was never meant to stay at the level of mechanical execution. We’re built to operate at the level of purpose, direction, and relationship. Genesis 1:28 gives humanity dominion and stewardship — a mandate to cultivate, not just maintain (Genesis 1:28, ESV).

The master in the parable didn’t give talents so the servants could admire them or lock them away. He gave them to be invested — put to work — in ways that generated return.

For those of us building technology that serves the church, the return isn’t financial. It’s pastors freed from busywork to do the work they were called to. It’s 75 million monthly visitors encountering Scripture through a platform that keeps getting better because the product team has time to think. It’s churches stewarding every tool available — including AI — in service of the mission they’ve been given.

The talent is in front of us. What we do with it is a stewardship question.


Josh Read is Director of Product at HarperCollins Christian Publishing (Bible Gateway) and holds a doctorate in Strategic Organizational Leadership. He writes about AI, product leadership, and digital discipleship at drjoshuaread.com.

The Big Five Personality Traits: What Every Change Leader Needs to Know

Most change management initiatives fail. Not because the strategy was wrong or the technology didn’t work but because leaders treated a room full of unique human beings the exact same and expected they would all respond to change the same way.

They won’t. They never do.

After more than a decade leading product and organizational change, running platform modernizations, launching new products, and driving cross-functional alignment at companies serving millions of users, I’ve learned one thing: how people respond to change is largely predictable if you’re paying attention to the right signals.

The Big Five personality model gives you a scientific framework for doing exactly that. It won’t tell you everything about a person. But it will tell you enough to stop being surprised when your most methodical engineer resists a process change that your most curious designer is already championing at lunch.

This article breaks down what the Big Five actually is, what each trait means in practice, and how you can apply it specifically to change management no matter what your title is.


What Is the Big Five Personality Model?

The Big Five — also called the Five-Factor Model (FFM) or referred to by the acronym OCEAN — is the most empirically validated framework in personality psychology. This isn’t Myers-Briggs, which has significant reliability problems. This isn’t an astrology-adjacent personality quiz. The Big Five emerged from decades of peer-reviewed psychometric research and is the standard model used in academic personality research worldwide.

The framework originated from what researchers call the lexical hypothesis: the idea that the most important personality traits in any culture will inevitably be encoded in its language. In the 1930s, psychologists Gordon Allport and Henry Odbert catalogued over 4,500 English adjectives used to describe people (I love this idea and think it would be so interesting to have watched them come up with some of the silly ones). From there, researchers spent decades using factor analysis, a statistical technique for identifying patterns in data, to cluster those descriptors into broader dimensions.

By the 1960s, Warren Norman had consolidated the research into five robust factors. In the 1980s, Paul Costa Jr. and Robert McCrae developed the NEO Personality Inventory, which remains the most widely used Big Five assessment today.

The key distinction from other personality models: the Big Five measures traits on a continuous spectrum, not binary categories. You’re not an “introvert” or an “extravert” but you fall somewhere on a range. That nuance matters enormously when you’re managing real people through real change.

The five traits, spelled out by the OCEAN acronym, are: Openness to Experience, Conscientiousness, Extraversion, Agreeableness, and Neuroticism.

Let’s go through each one in detail.


O — Openness to Experience

What it measures: Creativity, intellectual curiosity, and willingness to engage with new ideas, abstract thinking, and novel experiences.

High scorers tend to be imaginative, intellectually adventurous, and genuinely excited by complexity. Low scorers tend to be practical, conventional, and preference-stable — they want proven methods over experimentation.

Neither end of the spectrum is “better.” High openness without the grounding of conscientiousness can produce someone who loves ideas but ships nothing. Low openness combined with high conscientiousness can produce your most reliable operators.

In change management: High-openness people are your early adopters. They’ll be intrigued by the change, ask the interesting questions, and often become your internal champions. I recently had a large launch and my main champion was so excited that she had been using the prototype before I had even let some of my development team know about the project. The risk with high-openness individuals is that they get excited about the idea of change and then lose interest during the unglamorous implementation phase. I kept my champion excited by only introducing small features to her throughout the build.

Low-openness individuals will resist change more instinctively, not because they’re obstinate, but because they’re wired to value stability and proven approaches. What they need from you is not enthusiasm, it’s evidence. Show them data, precedent, and a clear picture of what stays the same. They’re not your enemy in a change initiative. Ignored, they become your loudest critics. Engaged correctly, they become your quality control. This guy is on my team too. He’s difficult to deal with if I don’t give him the data, but I’ve looked at it as a win because it does force me to slow down and get the data. I know he’s going to be asking me questions and I try to have the answers before he arrives.

The practical play: Segment your communication strategy. Don’t send one change announcement to your entire organization and assume it lands equally. High-openness stakeholders want to be in the room early, contributing to the shape of the change. Low-openness stakeholders want to see the pilot results before they commit. Build your rollout timeline to accommodate both.


C — Conscientiousness

What it measures: Self-discipline, organization, dependability, and goal-directedness. Essentially: how reliably does this person follow through?

High scorers are methodical, thorough, and tend to plan ahead. They show up on time, deliver on commitments, and prefer structured environments. Low scorers are more spontaneous, flexible, and sometimes prone to procrastination — but also often more adaptable in chaotic environments.

In change management: High-conscientiousness people are your implementation backbone. Once they buy into a change, they will execute it more reliably than anyone else in the room. The challenge is that they need the full picture before they commit. Launch an initiative without a detailed plan and they’ll spend the entire kickoff meeting asking questions that feel obstructionist but are actually their brain trying to build the mental model they need to be effective.

Low-conscientiousness people adapt to change more easily in terms of mindset but more erratically in terms of execution. They’re comfortable with ambiguity, which is valuable in early-stage change, but they need more structural accountability to follow through on the consistent behaviors that make change actually stick.

The practical play: During change planning, put your high-conscientiousness people in charge of the process design. Give them the runway to build the playbook. During rollout, give them clear milestones and let them hold the team accountable — they’ll do it naturally. For low-conscientiousness team members, build in more check-in touchpoints and make progress visible publicly. Not as surveillance, but as external structure that replaces the internal structure they don’t naturally have.


E — Extraversion

What it measures: Energy drawn from social interaction, assertiveness, tendency toward positive emotion, and degree of talkativeness and engagement in group settings.

This is the most commonly misunderstood trait. Introversion is not shyness. Extraversion is not confidence. The dimension is about where you draw energy from, social engagement energizes extraverts and drains introverts, not the other way around.

High extraverts are vocal, enthusiastic in group settings, and often the first to speak up. High introverts process internally, may need more time before contributing, and tend to prefer one-on-one conversations over all-hands forums.

In change management: Extraverts will often create the social momentum a change initiative needs. They talk about it at lunch, they advocate in meetings, they make it feel like something real is happening. That’s enormously valuable — but it can also mean they’re generating buzz ahead of the evidence, which creates a credibility gap if the initiative stumbles.

Introverts are processing the change deeply but quietly. The danger is confusing their silence with acceptance or indifference when they might be holding substantive concerns that never surface in a town hall setting. Some of the sharpest change-resistant thinking I’ve encountered came from introverted team members who raised it six weeks later in a one-on-one after everyone thought we were past the debate phase.

The practical play: Don’t let all-hands meetings be your only feedback channel. They structurally favor extraverts. Build in async feedback mechanisms — surveys, Slack threads with explicit prompts, written pre-reads before meetings — that give introverts the processing space they need. And deliberately seek out your quiet team members one-on-one before major decisions close. You’ll learn things you never would have heard in a group setting.


A — Agreeableness

What it measures: Cooperativeness, empathy, trust, and prioritization of social harmony. How inclined is this person to put others’ needs above their own?

High agreeableness correlates with warmth, generosity, and conflict avoidance. Low agreeableness correlates with competitiveness, skepticism, and willingness to challenge or confront. Confronting doesn’t necessarily look like hostility, but there may be a higher tolerance for tension.

In change management: This is the trait that creates the most dangerous blind spots for change leaders. High-agreeableness individuals will often nod along with a change initiative in a group setting even when they have legitimate reservations because disagreeing publicly feels like a social cost they’d rather avoid. When you see heads nodding around the conference table, do not assume buy-in.

Low-agreeableness individuals will push back openly and sometimes aggressively. That can feel uncomfortable and even disrespectful in a change rollout. One thing I’ve learned to appreciate (as alluded to above) is the person who tells you directly that your change plan has a hole in it, they are actually doing you a favor. The person who agrees in the meeting and then quietly undermines the initiative in the hallway is the real risk.

The practical play: Create explicit, structured opportunities for dissent. Not just “does anyone have concerns?” at the end of a packed meeting, but dedicated pre-mortem exercises, anonymous surveys, or devil’s advocate assignments that normalize pushback as part of good process. This gives high-agreeableness people a structured reason to voice concerns (it’s the process, not personal conflict) and channels the energy of low-agreeableness people productively into improving the plan rather than resisting it.


N — Neuroticism

What it measures: Emotional volatility, sensitivity to stress, and tendency toward negative emotional states like anxiety, irritability, and self-doubt.

High scorers experience stronger emotional reactions to uncertainty and stress. Low scorers (sometimes called emotionally stable) tend to remain calm under pressure and recover from setbacks more quickly.

This is the trait that demands the most careful handling in a leadership context. Neuroticism should not be viewed as a weakness or a character flaw. I do think there is a maturity level where it may appear as aspects of Neuroticism, but Neuroticism is a genuine dimension of human experience that shapes how people respond to threat and uncertainty. And organizational change is perceived as a threat by more of your team than you probably realize.

In change management: High-neuroticism team members will likely experience change as significantly more stressful than their low-neuroticism peers, even if the change is objectively positive. They need more frequent reassurance, more visibility into what’s coming, and more explicit communication that their role and contributions are valued. If they’re experiencing change anxiety and you’re not addressing it, it will express itself as resistance, absenteeism, or performance decline.

Low-neuroticism individuals may be so unfazed by change that they underestimate the impact it’s having on their teammates. If you’re a leader who naturally scores low in neuroticism, this is your blind spot. Just because you’re energized by the change doesn’t mean everyone else is. Watch for signals in your team: short tempers, declining work quality, missed deadlines during the transition period. These are symptoms, not character issues.

The practical play: Increase your communication frequency during change and especially about what’s NOT changing. The human brain in stress mode fills uncertainty with worst-case assumptions. Every uncertainty you don’t address explicitly will be addressed implicitly by anxiety. Over-communicate the stable elements: this team stays together, your role isn’t going anywhere, here’s exactly what the next 30 days look like. Predictability is medicine for high-neuroticism team members.


Putting OCEAN to Work: A Change Management Framework

Understanding the five traits individually is useful. Using them together is where the real leverage is.

Here’s how I’ve applied this thinking practically in change initiatives:

Start with team mapping, not org charts. Before you launch a change initiative, spend time mapping your key stakeholders across the OCEAN dimensions based on what you’ve observed. These aren’t formal assessments, but your working knowledge of how these people behave. Who asks the most questions before committing? (High conscientiousness / low openness.) Who nods in meetings but raises concerns later privately? (High agreeableness.) Who goes quiet during stressful periods? (High neuroticism.) This map tells you who needs what from you.

Design your communication strategy around the spectrum, not the average. Most change communications are written for the mythical average employee: moderately open, moderately agreeable, moderately stressed. That person doesn’t actually exist in your organization. Write for the ends of the spectrum. Give your low-openness people evidence and precedent. Give your high-neuroticism people explicit, frequent reassurance. Give your introverted team members async channels to process and respond. One message, multiple formats and frequencies.

Use personality diversity as a feature, not a bug. The knee-jerk response in change management is to minimize resistance. The better instinct is to channel it. High-conscientiousness skeptics will pressure-test your process and find the gaps before they become launch-day disasters. Low-agreeableness team members will surface the honest objections everyone else is thinking but not saying. High-neuroticism individuals will be acutely sensitive to signals that something is off and they’re often right. Build a change governance structure that treats these people as assets, not obstacles.

Match your change champions to the stakeholders they’re influencing. A high-extravert, high-openness champion will be brilliant at building excitement in your early adopters. They’re probably the wrong person to walk your most cautious, routine-oriented team members through the transition. Match your internal advocates to the personality profiles of the people they need to bring along. This is people strategy, not manipulation, it’s meeting people where they actually are.


A Note on Using This Responsibly

The Big Five is a lens, not a label. People are more complex than five dimensions, their trait expressions shift with context, and nobody should be reduced to their personality profile in a change initiative or anywhere else.

What this framework does is expand your range of hypotheses. Instead of concluding that your cautious colleague is “resistant to change”, which puts the problem on them and takes it out of your hands, you have a more useful hypothesis: they may be low-openness and high-conscientiousness. That means they need evidence and a clear plan, not enthusiasm. That’s a solvable problem. “Resistant to change” isn’t.

Used that way, the Big Five doesn’t box people in. It opens up more humane, more effective ways of leading them through the hardest parts of organizational life.


Frequently Asked Questions

What is the Big Five personality model?

The Big Five, also known as the Five-Factor Model (FFM) or OCEAN, is an empirically validated framework that measures human personality across five dimensions: Openness to Experience, Conscientiousness, Extraversion, Agreeableness, and Neuroticism. Unlike personality typing systems that sort people into categories, the Big Five measures each trait on a continuous scale, making it a more accurate representation of the actual range of human personality.

How does personality affect change management?

Personality traits significantly shape how individuals perceive, process, and respond to organizational change. People high in Neuroticism typically experience change as more stressful and need more frequent, explicit communication. People high in Openness tend to embrace change early but may disengage during implementation. People high in Conscientiousness need structured plans before committing. Understanding these tendencies allows change leaders to design communication strategies, rollout timelines, and feedback mechanisms that meet people where they actually are rather than where leaders wish they were.

Should you give your team a Big Five assessment?

Formal assessments can be useful, but they’re not required to apply this framework. Effective leaders can identify rough trait profiles through observation and noticing who asks detailed questions before committing, who speaks up readily in group settings versus one-on-one, who seems energized versus stressed by ambiguity. Formal assessments add precision; behavioral observation gets you most of the way there and doesn’t require an HR initiative to begin.

Is the Big Five better than Myers-Briggs for organizational use?

For research-backed applications, yes. The Big Five has substantially stronger empirical support than Myers-Briggs (MBTI), which has been criticized for low test-retest reliability, meaning people often get different type assignments when they retake it. The Big Five is the standard model in academic personality research precisely because of its consistency and predictive validity. For organizational change management specifically, the Big Five’s continuous scale model is also more practically useful because it captures the nuance that binary type systems miss.

What is the most important Big Five trait for a change leader to understand?

Neuroticism. Not because it’s more common, but because it’s most likely to be invisible until it creates problems. High-neuroticism team members experience genuine distress during organizational change that can express itself as resistance, disengagement, or performance decline if it’s not addressed. And leaders who naturally score low in neuroticism often don’t notice it in their team because they’re not experiencing the same stress response. Understanding and proactively addressing the anxiety dimension of change is one of the highest-leverage moves a change leader can make.

Schneiderman’s Eight Golden Rules

Ben Schneiderman worked in Human-Computer Interaction and his research revealed these eight golden rules for interface design.

  1. Strive for consistency. Use familiar icons, colors, menu styles, calls to action, etc.
  2. Enable users to use shortcuts. Users who use your product often will inevitably understand it and no longer need directions on how to use it. They will start looking for ways to move through the interface quicker, provide them shortcuts.
  3. Offer informative feedback. Breadcrumbs and ripple effects on websites, ATM noises, haptic responses on phones/watches are examples of informative feedback.
  4. Design dialogue to yield closure. Thank you messages after purchase, Congratulations after sign-ups, these messages close the interaction for the user.
  5. Offer simple error handling. This reminds me of back in the day when forms were really hard to develop and if you filled one out incorrectly, you would lose all of your information when the page kicked you back. Simple error handling flags fields that may have been missed or filled out improperly.
  6. Permit easy reversal of actions. If the user feels comfortable that errors are reversible, they will explore more.
  7. Support internal locus of control. If your users explore more, they will feel more in control and ultimately trust your application or company more.
  8. Reduce short-term memory load. Human attention is limited. We are only able to remember five things at a time (give or take 2). Recognition is always easier than recalling something.

Deep Dive Resources:

https://developer.apple.com/design/human-interface-guidelines/

This post is part of a series of quick informative lists I can refer back to when doing research or preparing presentations.

Information Facts of Life

According to an article from HBR (March-April 1994), there are rules governing information sharing behavior. Having run across these rules doing some Change Management research this morning, I find these rules relevant even 26 years later.

  • Most of the information in organizations – and most of the information people really care about – is not on computers.
  • Managers prefer to get information from people rather than computers; people add value to raw information by interpreting it and adding context.
  • The more complex and detailed an information management approach, the less likely it is to change anyone’s behavior.
  • All information does not have to be common; an element of flexibility and disorder is desirable.
  • The more a company knows and cares about its core business area, the less likely employees will be to agree on a common definition of it.
  • If information is power and money, people will not share it easily.
  • The willingness of individuals to use a specified information format is directly proportional to how much they have participated in defining it, or trust other who did.
  • To make the most of electronic communications, employees must first learn to communicate face to face.
  • Since people are important sources and integrators of information, any maps of information should include people.
  • There is no such thing as information overload; if information is really useful our appetite for it is insatiable.

Original Article can be found here.