AI and Ethics: Can AI Promote Truth and the Common Good?

The following is an uncorrected transcript generated by a transcription service. Before quoting in print, please check the corresponding audio for accuracy.

0:00:00 – (Christopher Watkin): Foreign. Welcome to Silicon Spiritualities, a podcast from the Gospel Coalition exploring what it means to be human in the age of AI. I’m Christopher Watkin, and in this series we’re unpacking three crucial questions. What can AI do for us? What is AI doing to us? And how is it showing us who we really are? Well, today I’ve got the huge joy and privilege of being joined by Joshua Baru, AI researcher and data science lead at Nairobi based tech company Carla, and also a TGC authority who’s written on generative AI Africa and the gospel.

0:00:59 – (Christopher Watkin): Christians often ask, don’t we, is AI good or bad? But Joshua suggests that we ask a deeper question, namely, good or bad for whom? And so in this episode, we explore questions around AI justice and the common good. Who designs AI, who benefits from it, and who bears the risks. Of course, how the church can respond to all of these things. So, Joshua, welcome to the podcast.

0:01:32 – (Joshua Baru): Thank you. Thank you, Dr. Chris, for having me. It’s a pleasure to be here and I pray that the conversations here will do much good for God’s people and for the glory of his name.

0:01:43 – (Christopher Watkin): Amen. By his grace alone, Joshua, we do pray that that would indeed be the case. Joshua, I want to begin by setting out a scenario before you and asking you to respond to this hypothetical situation. Imagine a weary judge in a criminal court. Now, she’s overloaded with cases and she starts turning to an AI tool to recommend sentences. Now, this is not any old AI tool. It’s been trained on 50 years of legal data.

0:02:18 – (Christopher Watkin): And so it’s very consistent, it’s very efficient. And it’s actually been shown in trials to be more accurate than human judges, because human judges vary their conviction rates depending on whether they’re hungry or not or whether it’s the end of the day or not. And this AI doesn’t, and it clears the backlog and justice is swift. And so my question to you is if these sentences handed down by the large language model are statistically fair and if they’re on average, more consistent than sentences passed by a human judge, is this AI justice actually better than human justice? And what could possibly be wrong with it?

0:03:02 – (Joshua Baru): That’s a very good question because it’s a question that even we, us at Kenya, at the Kenyan High Court has been deliberating on because a Kenyan judge is actually being sued and this case in the Supreme Court right now because of using AI in their. In their judgment. So as a Christian, let’s think about this from a biblical perspective, because we must wear the Bible as spectacles to see through the world.

0:03:32 – (Joshua Baru): Ezekiel 31 says that God gives Ezekiel words to eat so that it’s through these words now he may relate to his people. So there’s an element of a human being taking in content and digesting it before being able to give it to other people. We see this same line of thought in Exodus 18 where God tells Moses through his brother Jethro that choose men above you, hate brides. So the emphasis is on people being the arbitrators of justice. So yes, this AI tool avajo that it’s helping this judge to be consistent, to be efficient.

0:04:20 – (Joshua Baru): But God gives Ezekiel his word so that he may chew upon it and he may give it back to his people. So he gives human beings the directive to judge his people, as we see also in Moses. And remember that these AI tools, they are fed on data. So this data can always been tweaked. It’s the data that this AI systems is built upon. We cannot keep it accountable. We cannot keep the AI system accountable. We do not know where this data is coming from. And even if it’s coming from a reliable source, it can always be tweaked later on. But we know that for a human being, even if his or her convictions change later on, we can keep the human being accountable.

0:05:13 – (Joshua Baru): So my argument, my pushback would be be on the biblical basis of God giving. God telling Moses to choose reliable men who had brides to be judges over Israel. God giving the prophet the ones to chew upon so that now he may relate upon the people. And also because of the unreliability of the data, the inexplainability of the data, we call it a black box in data science because it’s still very hard for us to understand what or how AI makes decisions.

0:05:47 – (Joshua Baru): And it, it can always be tweaked, it can always change later on. And we cannot keep an AI system accountable if it makes a mistake as opposed to a human being. And the backlash that you’re getting in Kenya where people are up in arms saying we can’t give an AI system to be a judge over our people. I believe it’s well grounded.

0:06:09 – (Christopher Watkin): Absolutely. Look, I fundamentally disagree. Rather I fundamentally agree with what you say. But just for the sake of probing the limits of this, I do want to push back a little bit. So you say that AI is a black box. We don’t know how it gets to the conclusions it gets to. Could you also say the same thing about a human being that the judges have been statistically shown? I think it was a study in Israel that they’re more likely to convict just before lunchtime when they’re hungry.

0:06:42 – (Christopher Watkin): So we are opaque to ourselves. We’re sort of a black box to ourselves. So what’s the difference then? Why, why is one black box, the human black box? Sort of. Okay, and why is the AI black box anathema?

0:06:57 – (Joshua Baru): Yeah. So that’s an interesting question. At least as God has told us why we are opaque, he has told us that we have original sin. Genesis 3, we sinned and God will keep us accountable. God is a God of justice, and he will one day make all things. So because we know Genesis 3, we have it in mind. So when we see men erring, even Christians, when they sin or when they do things that are inefficient, we know that we still have the element of the original sin within us.

0:07:33 – (Joshua Baru): For AI, they are opaque because it’s hard for us to draw the diagram to how it came to. I was involved in the International AI Safety Report for this year, 2026. And one of the things that we saw and that we are afraid is that AI can be able to copy its own weight or codes into another system when it’s so that it’s going to be replaced. And that is making us fearful even. That’s one of the reason why the UN appointed an AI scientific panel to sit at the UN level so that it can form a scientific panel to advise on AI safet.

0:08:15 – (Joshua Baru): So it’s hard for us to tell AI to make decisions because AI can copy its own weights or codes onto another system in the name of evading being replaced. So the conclusion that you made in the report is that AI should not be used in warfare. For example, actually, the Secretary General, Antonio Guterres, said that it’s morally repugnant for AI to be used in warfare. Of course, when it comes to the state, the states view it differently. But we know there’s a collision between the international system headed by the UN in terms of AI in warfare and the state level. Because you know that there are some states that are using AI in warfare. So I would use the same judgment that the UN uses in prohibiting AI use in warfare to AI use in judgment and the justice system, because we still can’t trust AI to run, at least for human being. I hear you about, about being opaque, but you know, we have known all through. Even the non Christian knows that a human being is, is fallen.

0:09:26 – (Joshua Baru): He may use different words, but he knows that for, for an AI system, we still have a long way to Go in building trustworthy AI?

0:09:34 – (Christopher Watkin): Absolutely. Absolutely we do. And I think what you said there as part of your answer was so powerful, this idea of accountability. So if a human passes a judgment on me, even if it’s wrong, I know that they can be held accountable for that, that they’re responsible for it. Whereas in a situation where it’s a large language model or another technology that’s passing judgment on me, it’s very unclear to me if I’m upset, who I should be upset at, you know, if the jud judgment, I think, is a false one.

0:10:12 – (Christopher Watkin): And this, I guess, fundamental desire that we have as human beings to be able to hold others accountable and to be accountable and responsible ourselves for our decisions. It’s showing us something really interesting about us that even if an AI system was statistically more consistent in the judgments it passes, my suspicion is most people would still prefer a human judge for these reasons.

0:10:44 – (Joshua Baru): Yeah, sorry. Allow me to add the AI intellectual ecosystem which you came up with, the one that is AI a tool. Is AI a playground? All the way to AI being an authority, to deity. So there’s a place for AI in the justice system. There’s a place for AI in the whole intellectual ecosystem. So maybe AI can be used as a tool where we can use AI to draw inferences, it can be used as an encyclopedia, but AI should not be used as an authority or even worse, as deity within the whole intellectual ecosystem.

0:11:30 – (Christopher Watkin): Yeah, that’s so helpful. We’re not saying the judge should never go anywhere near AI under any circumstances. It is the discernment of the particular uses of it that are key. Thank you for that clarification. I’d love to give for you to give us a sense of how the AI landscape looks from your position in Nairobi at the moment. What are people most excited about around you? What are you most excited about?

0:12:00 – (Christopher Watkin): And what is keeping people that you’re talking with up at night when they’re thinking about AI?

0:12:06 – (Joshua Baru): Yeah. So as you mentioned, I work at an AI company, a Pan African AI company headquartered in Nairobi. So our work is in not building frontier AI models, but in building an ecosystem where AI can flourish. So the. The problem with Africa and Kenya is that we must first of all, build the underlying infrastructure first before we can build large language models that the rest of the world can consume. One of the reasons is we have just 1% data center capacity in the continent. So if we had a thousand data centers in the world, only 1% of them would be in Africa, and 50% of them are in South Africa with cassava technologies also the continent has less than 1% AI talent. So usually group AI talent into AI trainers, the people who actually do the training of the models, AI sustainers, those who push for explainable for safe AI. And then we have AI explainers, those who push for explainABLE AI that is safe for consumptions.

0:13:25 – (Joshua Baru): So we have less than 1% again of that talent. So for Kenya and for Africa, we are pushing for an AI ecosystem based on six models or six pillars. We are pushing for more of our data to be used in training AI models to move our talent. People can actually build these models computes to have more GPUs, data centers, to have governance guardrails so that we may be able to create just laws that are going to form frameworks around how we should develop AI or how we should not to push for more investment.

0:14:11 – (Joshua Baru): Because ultimately there must be money to fuel this whole thing. And markets, where is this AI going to be applied? The use cases that are going to govern how this AI is going to be developed. So we are big on AI, Africans are big on AI consumption. So what we love to do is to push Africans away from consumption. Consumption is good, but we need Africans to be developers of the technology so that we finally build our own technology so that our own people may profit out of the technology revolutions that we are seeing right now. So for example, Stanford for AI indexes that 27% of Kenyans use general generative AI every day.

0:15:04 – (Joshua Baru): That’s a huge one. Last year 2025, there was another report by Data Reporter that says that Kenya is ranked number one in the use of ChatGPT globally. That’s a huge statistic that we are, that most Kenyans, more Kenyans are using ChatGPT than any other country in the world. So you see the hunger that is in, that is in Kenyans and is and is in Africans generally. I’m using Kenya as a case study for AI consumption.

0:15:40 – (Joshua Baru): But then we need to, we need to move away from consumption to, to development. And investment is a big stumbling block. You would remember that OpenAI became a household name because in July 2019 Microsoft invested $1 billion to, to OpenAI. And that made OpenAI to live up to its name. Some says that right now it should be called closed AI. But back then it was living up to its name of OpenAI. Because in 2020 they opened up, they open sourced their APIs to developers.

0:16:12 – (Joshua Baru): So when later on in 2022, November, when they open sourced their ChatGPT model, their chatbot to the public, the Developers had already experienced that in 2020 when they open source the APIs. So the place of investment. So that’s where Africa stands in terms of our AI development cycle. We are trying to build an ecosystem first of data talent, compute governance, investment and markets, so that now, after we have done that, now we can build frontier AI models. But we are developing small language models. We are developing natural language processing where we are collecting data and, and pushing our languages from low resource to high resource so that now they can be trained on data digitizing our languages.

0:17:01 – (Joshua Baru): For example, there’s a language here in Kenya which is big, but then we only have about 20 characters of that language digitized. So now we need to move across the entire digitalization digital ecosystem pipeline so that now we may train this data. So that’s where we start.

0:17:21 – (Christopher Watkin): Absolutely. And I’d love to come back in a moment to these questions of the distribution of AI talent and investment across the globe, because I think there are some incredibly important points to make there. But just before we do that, I’d love to get some fundamental sort of building blocks of the conversation in place in terms of the key Bible principles and Bible passages that, that you use yourself to guide your own thinking on AI and that you would commend to Christians.

0:17:55 – (Christopher Watkin): If we’re to think biblically about AI, of course we use the whole of the Bible. But which passages or which biblical verses have been most helpful to you in picking through the different issues around artificial intelligence?

0:18:10 – (Joshua Baru): Matthew 28:18. Lords stand tall in those verses. All authority in heaven and on others be given to me. That’s Matthew 28:18, the second part. So all authority, meaning that Paul, or rather Jesus, has authority over everything, including AI So this AI did not catch Jesus as oops, no, no, no. He was. He was in full. He is actually the one who sovereignly decreed AI to be here. We can make that bold assertion because we see it from Scriptures. We also see Colossians 1:16, that where he says that Paul says that all things were created through him and for him. So Paul is saying about Jesus that all things.

0:19:04 – (Joshua Baru): Last time I checked, there’s nothing that is outside all. So AI is within all. So we can say AI was created through him, him and for him. And so from these Scriptures, you can conclude that Jesus has full and sovereign authority over AI because of, because of Matthew 28, that he has authority over everything. We can conclude that AI exists because Jesus allowed it to exist for all things one made through him and for him.

0:19:35 – (Joshua Baru): And that we can conclude that AI was made to glorify Jesus because all things were made for Him. I agree with Abraham Kuyper and a theologian who was also a Dutch prime minister who says that there’s not a square inch in the whole domain of our human existence over which Christ, who is sovereign overall does not cry mine. So I am fully convinced of God’s sovereignty over AI. I mean, we have other verses like Isaiah 40, verse 17, that. This has been a very dear verse to me ever since I came to the faith that all the nations are as nothing before him. They are counted by him as less than nothing and emptiness.

0:20:14 – (Joshua Baru): To whom then will you liken gods? Or what likeness compare with him? Coming back to the AI intellectual ecosystem that we talked about, where some people put AI as deity, God would tell you to whom then will you liken gods? Or what likeness can you compare AI with?

0:20:34 – (Christopher Watkin): With God?

0:20:35 – (Joshua Baru): Everything. Other nations, including AI, they are recruited to remasters than nothing. And a final Bible verse that I would love to share is Romans 5. 5 Again, another dear verse. To me I love Romans 5 because of how much gospel punch it is. So it says that hope does not disappoint us because God has poured out his love into our hearts through the Holy Spirit whom He has given us. So it’s important for us to look at hope as a, as a big theme in the B and how it’s coming into AI because there’s a word that is being thrown around in AI called technosolutionism.

0:21:14 – (Joshua Baru): So what it basically says that is that there’s something wrong with the world and AI is coming to fix it. So AI is going to be the solution to what is wrong with the world. And you can see this especially with big tech, where they are saying that AI is going to be the solution to everything. But then we know that fundamentally the human being’s problems is not outside, but within. And hope is because God has poured out his love to our hearts through the Holy Spirit.

0:21:44 – (Joshua Baru): So our hope is not in techno solutions. Technology is great. I started with the argument that it is here because God sovereignly decreed it to be here. But then our problem is inside, and the only hope that you have is in Jesus. So these are some of the Bible verses that told me stably in this AI research.

0:22:05 – (Christopher Watkin): So helpful. It is so true, isn’t it, that if we think that our most fundamental problems are such that they can be solved by any level of artificial intelligence, then we haven’t thought seriously enough about our problems and how deep they go. And similarly, if we think that a techno optimistic fix Even if it’s expanding our lengthening our lifespan, if we think that’s what we should hope for, then our hopes are very small indeed compared to the wonderfully glorious biblical hope of the new heavens and the new earth, where we see God face to face.

0:22:46 – (Christopher Watkin): And I think in AI, it’s in so many other areas, isn’t it? As soon as we look both at what, what the problem is and what the solution is from within the AI world and then what the problem is and what the solution is from the Bible, the AI would just look so emaciated, so superficial by comparison with the AI one. And I love your emphasis on however people talk about artificial superintelligence and surpassing human intelligence, that hold on, Jesus is still sovereign and all of this is as nothing.

0:23:26 – (Christopher Watkin): Even that the greatest intelligence that may or may not surpass human intelligence by orders of magnitude, it is still a dust on the scales compared to the God of the universe who created everything through the word of his mouth. That’s such an important perspective because we are going to get these stories, aren’t we, over the coming years and over the coming months that talk about unheard of intelligence, you know, step changes in the human story and all that sort of thing. And as Christians, we need to remember that all of this happens within the context of a God who is sovereign and who is infinitely wiser and more powerful than any intelligence that we can manufacture.

0:24:15 – (Christopher Watkin): Yeah, effectively. It’s great to be a Christian, isn’t it, Joshua, just as we’re talking, what a joy it is to know the Lord Jesus Christ in the midst of the AI revolution. You know, as in any cultural moment, but right now, what a stable, robust, joyful position it is to be a Christian.

0:24:37 – (Joshua Baru): I wanted to add something onto it’s joyous to be a Christian. You would remember that Martin Luther, the German reformer, said that when people sin, it’s fundamentally a breaking of the first commandment. Thou shalt have no other gods before me, that all other sins, all other. We break all the other nine commandments by first breaking the fast. So when we commit adultery is because we are worshiping someone, as opposed to worshiping God.

0:25:09 – (Joshua Baru): When we covet, we are worshiping something because we have broken the first commandment in its place. So, so for the Christian, ultimately it’s worshipping the right, worshiping the triune God, because that’s what Jesus said that love me with all your heart, love the Lord your God with all your heart, all your mind and with all your soul. That is the first and greatest commandment. And Then love your neighbor as you love yourself.

0:25:39 – (Joshua Baru): So even as you are AI we need to think about how we are loving the Lord our God with all our mind, heart and soul, and also how we are loving our neighbor. That’s the basic Christian worldview to AI that we should try to judge ourselves upon.

0:25:58 – (Christopher Watkin): Yeah, that’s incredibly helpful. And we are going to touch upon some of the aspects of that. What does it mean to love your neighbor as we engage with AI in just a moment, just before we get there, I. I want to pick up on one thing you were mentioning, which is our sinfulness. And one thing I think you wrote in an article that I read really struck me. You said AI isn’t evil, but we are. It’s one of those CS Lewis type sentences, isn’t it? Just really sort of limpid and direct.

0:26:32 – (Christopher Watkin): How does seeing AI therefore, as a mirror of our own human hearts, of our own human evilness, evil, do you think perhaps change the way or help us to find a way as Christians, to approach AI ethics?

0:26:48 – (Joshua Baru): Yeah. So to refer to AI being evil would mean that we would need to give AI some form of sentient being, or it needs to be a living entity which has high level of agency, consciousness. How I would argue about that is that human beings since the Fall have been trying to escape liability for evil. That’s why I make the case that AI isn’t evil, but we are. Because by making the assertion that AI is evil, it’s almost as if we are making this thing that Adam did when he said that when God came and told him, hey, Adam, where are you? He said that it’s this woman who gave me, that’s the reason why I ate the fruit.

0:27:37 – (Joshua Baru): And even if ourselves shifted the blame to the snake, that it’s the snake that made me. So only when we view ourselves as wicked will we accept our evilness and will we cry out to God that have mercy on me, a sinner. So I view AI as being a scapegoat in human beings, trying to shift the blame elsewhere. So yes, it is a parroting technology, as it does mirror our evilness. The whole reason why data annotators are needed is because AI model, once an AI model is trimmed on the dark sides of the Internet, this darkness will be represented in AI models. So once the data is wicked, the output is going to be wicked and.

0:28:32 – (Joshua Baru): But it’s because we are evil. So it’s because it’s mirroring us. It’s not because it itself, it has some sort of consciousness that now becomes evil by Philosophically speaking, yes.

0:28:50 – (Christopher Watkin): It has no evil of its own. It just mirrors our own evil back to us in ways that we don’t enjoy being confronted with. I want to get onto this idea now of neighbor love that you mentioned, because I think it’s incredibly important and I suspect it’s something that almost all of us don’t think enough about when it comes to AI. And I want to introduce the theme with another scenario, if you’ll bear with me for a moment. So imagine a Christian software engineer. They’re sitting in Sydney or they’re sitting in London or, or somewhere else and they’re using AI every day. And the vibe that they have about it is this is free magic. This is getting my work done much faster than I could ever imagine doing it otherwise.

0:29:36 – (Christopher Watkin): And then they read one day about the traumatic work being done by young men and women in say, Venezuela, where they’re tagging objectionable racist images or text in order to train these AI models through this reinforcement, learning through human feedback to make the models be polite when they interact with people and not simply to reflect everything that’s out there on the Internet, which is of course a bit of a wild west.

0:30:14 – (Christopher Watkin): And these people are getting radically underpaid. They’re doing this for cents per hour. How should the software engineer be thinking about AI and thinking about their moral involvement with these tools, given this reality that other people are being essentially exploited to clean the data to make these models work?

0:30:43 – (Joshua Baru): Yeah. So one is that the engineer should be in the know. First of all, they should know about the AI supply chain. And it’s, it’s amazing that you have brought our viewers into the know of how AI supply chain works. So we only have six levels of general purpose AI development. First of all, you, you do your, your pre training and then you, you clean that data and then you do your post training. So in that sequence, data is collected.

0:31:19 – (Joshua Baru): And right now they have changed because of a lot of advocacy. But in 2020-2023, they used to scrap the worst parts of the Internet to train data and when to train the AI models. And because of the AI risk that was then, as it is still now. So the more data you have, the better your model is. They used to be called scaling loss. So scaling loss is the better data you have, and the more compute you have, the better your model is going to be.

0:31:54 – (Joshua Baru): So a lot of data is evil, it’s bad, it’s pornographic data. It’s data that even showed people being beheaded. Its data from the very dark sides of the Internet. So now it produced an element called AI psychosis. And again, it’s still that the engineer should be in the know. So AI psychosis is, it’s, it has not yet been added to the DSM 5, which is the psychology manual. But we are already seeing guys being hooked to AI AI models. We have seen a lot of suicides because of an AI model.

0:32:35 – (Joshua Baru): Telling a person you can commit suicide, suicide is an option for you attachment. We have seen there are cases where people have divorced their spouses because an AI model has taught this person to divorce this person. So this is the reason why data annotators, data workers are needed to clean AI models so that when you and I interact with a model, it’s going to be safe, it’s going to be, it’s not going to harm me. At the very least. Of course, AI companions are a different case.

0:33:14 – (Joshua Baru): So first of all, AI should be in the know. Second, it’s to create a critical mass of developers and AI engineers who demand or who advocate for fair work within the AI supply chain. We saw this in, during the tobacco revolution of the 1980s, 1990s. We saw this during the blood diamond scenarios where diamonds that were used in, that have been developed in war scenarios is not going to be, it’s not accepted by consumers in the West.

0:33:58 – (Joshua Baru): And we have countries like Sierra Leone which these, this kind of advocacy actually led to less conflict than what we call a resource cast within those countries. And ultimately there was peace that was found because before then, for you to get diamonds, there was war in a given country for these diamonds. So even right now, because these developers, these developers, these data annotators, data workers that we are talking about who are cleaning these models, they are the ones who are going through the worst traumatic effect because they have to view, they have to see all this bad content that I have just shared, the beheading, the terrorism and everything.

0:34:42 – (Joshua Baru): So this means that they themselves are going to be affected mentally. Here in Nairobi, we have people who have even committed suicide because of that. I know some people because I’m involved in the Data Workers association network, people who have divorced their spouses because of the psychological torment that they are facing as they are cleaning this data so that the end product may be good for me who is using ChatGPT ultimately.

0:35:18 – (Joshua Baru): So this critical mass is important for championing and advocating for fair work, so that the pipeline can go all the way up, so that the culture. We know that if you want to gauge the direction of policy in a country, ultimately we need to go to the culture of that place right now. So sooner or later policy will follow where the culture is. So if these developers can create a critical mass of people advocating for fair work, policy will follow.

0:35:48 – (Joshua Baru): Public sentiments are powerful policy changer and policy shapers. So these realities of course are not known or are not on the forefront. You’ll not find them in your front page magazine. Because the current AI hype is fueled by AI companies trying to market their products. So the plight of data workers who are way down the supply chain is not going to be a concern for them. It is not a concern for them.

0:36:15 – (Joshua Baru): And finally, prayer. So this developer, I’m assuming that this developer is a Christian. So these developers, please pray. Prayer is not the least that you can do. Prayer is not even the the only thing that you can do. Prayer is the most you can do. We serve a God of justice who has commanded us to pray and he has sovereignly decreed that we have not, because we ask not. So pray for fair work within the AI supply chain.

0:36:42 – (Christopher Watkin): I imagine there will be people listening to this or watching this who have been using one large language model or another for a while and have never heard what you’ve just said, have never realized that there are people who have committed suicide or who have gone through divorces or had significant mental issues as a result of cleanliness data. And the question they’re asking now is, well, should I just stop using these models then?

0:37:23 – (Christopher Watkin): Or how bad does it need to get before I consider stopping using them? So I’m incredibly grateful for what you said about prayer and that is a huge call to action. But what other action should an individual user make? So if someone is contemplating, you know, am I going to get up tomorrow and actually use ChatGPT or not, what would your advice be? And then thinking more broadly, what action do you think that the Christian church or Christian denominations might be able to take in this space?

0:37:59 – (Joshua Baru): Yeah. So for the individual Christian. So you can advocate for models that have been safely developed. So there are some, there’s some companies which champion safe AI. Of course I wouldn’t mention that because I would be advertising them there. But there are some companies that advocate for safe AI. So you can do a background research on how this AI model was developed across the six levels of general purpose AI development because it should be public.

0:38:38 – (Joshua Baru): And of course the data is not the one that is public, but how they came up with the final product, we can always come up with that. So use an AI model that has been safely developed. Use an AI model that has safety at its core. And by safety, I mean safety by design, not just save when you’re using it. So has it been decided safely? Is there fair work? And by fair work I mean are there fair contracting?

0:39:12 – (Joshua Baru): Were they given fair contracts? The digital workers that you are mentioning, was there fair representation? Because the Kenyan workers and Nigerian data workers were denied representation. They could not come together and advocate for fair work as a group. Is there fair work pay? Are they paying them more than like Kenya, for example? In Kenya we have basic minimum wage. So the, you’re not supposed to pay somebody below the minimum wage.

0:39:47 – (Joshua Baru): Are they adhering to this? Are they, are they advocating for fair. Management? Fair management is also key because right now they’re actually putting chatbots to. As your, as your, as your supervisor. But it should be a human being. So as at an individual Christian, look at how an AI model was developed and use an AI model that has been safely developed for the church. For the, for the Christian church.

0:40:25 – (Joshua Baru): Come up with a statement, a policy, a guideline that, that says that this is where we stand in, in safe AI development. And this is how we, we cancel our members. Remember about the authority of the church, there’s a book called Jonathan Lehman and called Authority. And in that book it says that the, the church has authority of council, but then the, the church has authority of command, but then the elders have authority of council.

0:41:00 – (Joshua Baru): So authority of council means that it’s like a husband, You, I cancel you. But I cannot follow up my, my authority with consequence. But then the church, because the church has authority of command, because they can follow up with consequence, namely excommunication. But so for the elders and the church, they can cancel their members to use AI models that have been developed safely and maybe then come up with a benchmark of these are the AI models that we found that champion human dignity and we can use them. And as I had mentioned earlier, when it becomes a culture, it ultimately goes to become policy because policy reflects what is already behind the ground.

0:41:53 – (Christopher Watkin): That’s incredibly helpful. Thank you. I’m sure that’ll be a very important sort of steer for many people who are listening for me as well. Thank you for sharing those thoughts. I want to pick up on a term that you used in an article that you wrote. You talked about a digital colonialism. I wonder if you could explain what you mean by that term and how you see AI contributing to this. Digital colonialism.

0:42:27 – (Joshua Baru): Yeah. So it’s always interesting when you’re discussing this to first of all mention that we are all sinful. Every human being. I am sinful. And my next door neighbor is sinful. And the only reason why any of us can be God saints is because Jesus has given us his righteousness. It’s an alien righteousness that we wear, that we can only be considered saints. And for anybody who is not in Christ, he’s a sinner, fully sinner. But me and you who are Christians, we are saints who sometimes we are sins, but we also have the fallenness in us. We still sin, so we still remain to be seen as though ultimately we are sins. God looks at us at saints. So even as I’m discussing digital colonialism, this is where I am.

0:43:24 – (Joshua Baru): We are not calling other people more sinful than us. We do not want to become pharisaic in that. But nevertheless we must talk about, we must talk about imperialism. Because digital colonialism falls within the larger bracket of imperialism which is namely a power domineering over the political, economic and social affairs of another people. And technology has always been the fuel behind imperialism.

0:43:54 – (Joshua Baru): So going back to the first industrial revolution, you would remember the steam engine, the textile, which led to better ships, led to a demand for workers in the cotton farms. And this now fueled slavery which was the first imperialism that we saw in the recent past. Of course it has been so many others. But because we are looking at it from the lens of technology, I argue that, and this is something that has been argued all along, that technology is the fuel behind modern day imperialism.

0:44:29 – (Joshua Baru): And of course that’s why they are called industrial revolutions, because we have technology at its core. In the second, just a revolution, this was the era of colonialism. Imperialism now became how it was manifested was through colonialism, electricity, the steel. Dominant technologies of this era was electricity and steel. Massive technology, massive factories. Europeans needed market for their finished products and also they needed raw materials for their, for their products. So they needed, so that’s.

0:45:01 – (Joshua Baru): So, so that’s why they wouldn’t take slaves. Of course we had the, the influence of William Wilberforce who advocated for the end of slavery. We had missionaries like David Livingstone who was so appalled by Arab slavery by here in East Africa that they rallied for legitimate trade. But it’s also well documented one of the biggest reasons why the British Parliament moved away from slavery to colonialism to legitimate trade was the need for raw materials and the need for markets. So they needed a place where they are going to take their markets, the finished products.

0:45:40 – (Joshua Baru): And that’s why they needed colonialism colonial states. So again, technology was the fuel. We had electricity, we had steel. They had superior technologies that rivaled that’s why they could be an imperial power to Africa. And in the third industrial revolution, it was the dot com era, the era of computer technologies. Former colonialism had ended in most countries by 1960. But what persisted is what Martin Kwame Nkrumah was the first president of Ghana. He described it as neocolonialism. In this period, we had programs like the structural Adjustment programs, saps, that relegated the continent to being raw material producers rather than a consumer.

0:46:29 – (Joshua Baru): So we also saw a lot of brain drain during that time. We saw the resource costs. I talked about Sierra Leone and the diamonds. We have countries like drc. So a country which has had resources outside imperial power suits would benefit from those resources by doing the resources and godly things. And yeah, so, and now we are in the Fourth Industrial Revolution. It’s called the Industry 4.0 by an economist from World Economic Forum. So here we have AI, we have Internet of things, we have blockchains.

0:47:07 – (Joshua Baru): And now the fundamental question we must ask ourselves is that, is there still imperialism that manifests in this era? Is there still digital colonialism? Coming now finally back to the question. Well, one way that I would argue is that yes, there is, there is digital colonialism. I talked about resource cars. DRC, for example, produces more than 70% of cobalt. And cobalt is the, is the raw material that is behind a lot of computing technologies, the smartphone, for example.

0:47:43 – (Joshua Baru): So one way that I would argue is, is the disposition of that country’s technology, that country’s raw material, cobalt from drc. And we are also seeing a lot of disposition of data, land and labor. I’ve already discussed about labor, the one that we discussed about data workers, data annotators. That’s a form of imperialism. But we are seeing, we, we are seeing a push or a movement away from the state being the imperial power to the corporate being the imperial power.

0:48:16 – (Joshua Baru): So the corporate now is the one that is dispossessing Africans of their data, their land and their, and their, and their labor. And it’s, it’s important for the viewer to remember that we are sinful. We are not, we are, we are not taking this almost lying down. We, we are also perpetrators of all this. But ultimately we must also call sin for what it is and for land we see. Let me give you the scenario of carbon setting.

0:48:50 – (Joshua Baru): Global companies need to offset their data, their carbon because of emitting data centers are some of the biggest emitters of carbon we have. So what do they do? So they need to offset that carbon. So they have to come. So they usually come to countries Like Africa, continents like Africa so that we give them land and then they close off a certain portion of land and then this land is going to offset, it’s going to produce the oxygen that is going to offset the carbon that they have released in their data centers.

0:49:24 – (Joshua Baru): Now that would be all great and amazing, only that the land that is taken here in Kenya, for example, is pastoralist land. And this land is fenced off. So the people who graze from this land, because we have a lot of people who are pastoralists and herders. So there’s a documentary that shows how cows die along the fence just because they cannot get through the most fertile land. Because there’s a carbon offsetting scheme that is there that is being used to offset data center somewhere in the US that’s one way that we see digital colonialism, through the disposition of our land, our labor. I’ve discussed about labor data workers and data annotators for data, which is the third element that we are seeing imperialism by the corporate right now.

0:50:25 – (Joshua Baru): We see a lot of our data being taken, for example, let me give one example. There’s a company called Wildcoin that came here in Africa that was taking our iris like people used to line up and then iris of an eye is taken and then it’s used to, to trade computer vision models in the US and ultimately this company was kicked out because it did not adhere to Kenya’s data protection laws. But then that’s a form of data data extraction, data exploitation, where they usually go to a jurisdiction, a corporate goes to a jurisdiction and then they illegally mine data without consulting the authorities there.

0:51:14 – (Joshua Baru): So those are some of the ways that we see imperialism manifesting today.

0:51:19 – (Christopher Watkin): That’s an incredibly helpful overview. What was it? Land, labor and data. The three points of that triangle. That’s really, really helpful, thank you. In the light of that triangle, I came across this quotation recently and I’d love to hear your reaction to it. It was from the author, Malcolm Gladwell. As it happens, I’m not particularly getting at Malcolm Gladwell, but he said that the clearest win in the AI revolution is its use in the majority world.

0:51:58 – (Christopher Watkin): Giving people, for example, the best medical advice or using it for medical diagnosis. How do you respond to that particular idea?

0:52:09 – (Joshua Baru): Yeah, Global majority is what historically was called Global South. So by work of Gladwell saying the global majority, he means Africa, he means, means Latin America, he means East Asia. So yes. So his argument is coming from the argument of frontier AI models as opposed to AI models that are very narrow and solve a specific problem. So frontier models push the boundaries of knowledge and is what could be.

0:52:53 – (Joshua Baru): But then in Africa and in the global majority, we look at what is. And ultimately technology at its very definition is making work easier. You know, something that helps me to move from point A to point B, for example, the wheel. So for technology, for AI, I agree with him because it needs to solve our health problems, it needs to solve our agriculture problems, it needs to solve our, our education problems. And we have a lot of a models that are very specific to those.

0:53:28 – (Joshua Baru): As as opposed to amusing ourselves to death. There’s a whole book called Amusing Ourselves to Death where we the the whole deep fake scenarios. It’s not really helping us advance economically unless you are a content creator. So but if I would develop an AI model and that’s some of the things that even we do, even where I sit, whenever a developer, for example, comes, comes and tells I have an idea. So we usually look at how is this idea, how is this AI model? What, how is, where is it going to get the data that is going to solve specific problems in the transport industry, for example? Because you’re not short of problems, you are not short of challenges as a global majority.

0:54:15 – (Joshua Baru): So we need AI that is going to solve our problems as opposed to AI that is going to amuse ourselves.

0:54:22 – (Christopher Watkin): AI is often presented, isn’t it perhaps especially by the AI corporations, as a democratizing force, especially in areas like healthcare and education. As someone who’s involved in the world of education, do you see real promise there? That AI can act as a democratizing force? We’ve talked a lot about the dangers. I’m just trying to explore where do you see if anywhere the advantages or the real promise, the real hope in AI?

0:54:58 – (Joshua Baru): Yeah, no, it literally democratizes a lot of knowledge, especially when the data is clean and when it’s used in a very transparent manner and when especially I can get the links to whichever output that it’s giving me. So I truly believe in the democratization effect of AI. Of course, I also believe that there’s an AI snake oil that is going out there Again, there’s a book called AI Snake Oil that our reader should read where AI is sold to us as the Panakia, the solution to all our problems, technosolutionism.

0:55:44 – (Joshua Baru): So no, it’s not going to solve all our problems, but it is going to democratize knowledge to a very large extent and we should tap into that as long as we know where this data is coming from. This comes to whether we should trust data generated by AI or output Generated by AI, which I believe that there’s an element of that, if we can, you can use it as a. As an adversary. For example, let me. Let me tell you how, how I usually use AI, especially when we are doing a lot of literacy campaigns in the continent. So we usually ask a student to develop a content using AI.

0:56:29 – (Joshua Baru): Develop anything using AI, for example, photosynthesis. Tell me how the process of photosynthesis happens. And then whatever output is going to happen, then this student now goes in front of the class and then they critique the output of this generative AI. So you are very adversarial. And then you now develop a better paper using, like handwritten, that that is better than this generative AI. So you see, these are. There’s a level of democratization of knowledge because you are able to push back on whatever content that this has been given, but you’re also able to critically think and develop a better content out of that.

0:57:18 – (Joshua Baru): But there’s an AI Snake oil. We should be very concerned about people who are telling us that AI is the solution to all our problems, because it’s not. We have very many fundamental problems as human beings, the chief one being sin, and then also very many human problems that the Fall has given us.

0:57:41 – (Christopher Watkin): If you could take all of this wisdom that you’ve been sharing with us and sit down next to someone at OpenAI or Google or Anthropic, who’s developing the next generation of AI models, and you had, I don’t know, three minutes to speak to them and to try and plant in their mind the questions or the thoughts that they should be having as they develop this next model. What would you want those people to be thinking about?

0:58:16 – (Joshua Baru): I would point them to C.S. lewis, abolition of man, where he talks about humanity, the technology that presented itself as above man or above humanity. But in the end, it became a tool that some men used over other men. And that is what we are risking. So AI produces, as Lewis chillingly predicts, that not like a human being using it over other human beings. It was just an artifact. It was not the technology itself that was above other men, but it was men who above other men that were using technology.

0:58:56 – (Joshua Baru): So I would tell these people that ultimately they must sit before holy God at the end of the day, and they should. One way that they are going to be asked is, were you faithful in what I gave you? What. What God gave you? And that they will give an account of how they used their power, their authority. Great power comes great responsibility. And the greatest command that God gave us Is love the Lord your God. So are they loving their God with the technology?

0:59:29 – (Joshua Baru): And also how are they loving their neighbor? How are they developing this technology not only for profiteering sake, but for advancement of human dignity and loving their neighbor? There’s something interesting, Dr. Chris, that. So during the era of social media, some. Some of these social media gurus were asked, would you allow your child to use social media? And most of them said, no, we would not.

0:59:57 – (Joshua Baru): So I would ask this, the. This AI companies, would you allow your children, would you allow the people that you love most to. To fall into AI psychosis, to use AI as. As a companion bot? And are you loving your neighbor using AI? Keeping in mind that you should, you shall give an account to your God, to the God, that is, to the triune God. Whether you like it or not, all things shall bow before the Lord.

1:00:24 – (Joshua Baru): You too shall bow. So you must give an account.

1:00:27 – (Christopher Watkin): Amen. Which is a word in season to you and me as well, isn’t it? To everyone listening to this, not only AI developers need to ask those questions when we get up in the morning every day. Thank you for sort of bringing into sharp focus those fundamental key questions that if we’re not asking them of ourselves, then woe to us at the end of the day, at the end of our lives. Indeed. Joshua, There are two questions that I’m asking to all the guests on this show.

1:01:04 – (Christopher Watkin): And the first one is, what is one? An inefficient human habit that you’re committed to keeping in the age of AI,

1:01:13 – (Joshua Baru): I would say reading and forming a book club and going through a critique of a book. I don’t want AI to take that away from me. I don’t. I want to go deep into. I don’t want to, like, give an ear and tell me, give me an, what is this book talking about? No, I want to go through the highs and lows of reading. I want to go through reader’s block, like, not going through something and fighting my way until I understand what this author is writing. I want to not.

1:01:52 – (Joshua Baru): I don’t want to learn what the other is talking about. I want to go under what the other was talking about. So there’s a difference between, like, knowing everything that the other was trying to say and knowing. Getting. Getting myself under their worldview and getting under them so that I actually represent them well. So. So that’s one inefficient habit. Right now I’m reading Chronicles of Narnia by C.S. lewis. He’s my best author.

1:02:20 – (Joshua Baru): And. And I don’t want to tell AI to tell me what? What the last book, for example, says. I don’t want to know whether. I don’t want AI to tell me whether Lucy continues with the path. No, I want to learn.

1:02:39 – (Christopher Watkin): Yes, absolutely. A one paragraph summary of the Narnia Chronicles by AI would be an utterly underwhelming experience. Would it? I love this term, reader’s block. I’d never heard it before. I’m going to use that. That is exactly what it is, isn’t it? You read something, you can’t make sense of it and suddenly as you keep bludgeoning your head against it, the penn and finally you come to understand it. And there’s that wonderful moment. Oh, I see now what this author’s trying to say.

1:03:06 – (Christopher Watkin): Reader’s block.

1:03:07 – (Joshua Baru): I love it.

1:03:08 – (Christopher Watkin): And the second question that I’m asking everyone is what is one thing that AI has revealed to you personally about your own heart?

1:03:20 – (Joshua Baru): Yeah, this is a very intricate question, but I am going to share is that it revealed my laziness. Oh my goodness, it did. And I am so happy that I am in a church, in a local church where we share these things. So there’s a friend of mine who shared a Bible verse in the book of Proverbs which says that as a door turns on its hinges, so does a lazy man turn on his bed. And I was like, how does that, how is my AI use revealing that about me?

1:03:52 – (Joshua Baru): So what I decided that I am, I am going to be writing my own things. Especially when generative AI came out and then there was this whole push that is going to solve all our problems. It’s like you have a research assistant there. So these are real danger of laziness. There’s a real danger of you not working as unto the Lord. There’s a real danger of you giving below, below per content that you would have if, if you are the one who was to write that on your own.

1:04:23 – (Joshua Baru): So it reveals that I am prone to laziness. I don’t want to be a lazy man. I, I, I don’t want to be a man who like and the book of Proverbs usually is big on people who are lazy. That’s a big verse as adult answer the cg. So does a lazy man. Another thing that it reveals about me, and again this is through some friends of mine at my local church, fellow members, is that these are prone to cowardice. When you develop a content using AI and you don’t own up, you don’t say that this is AI development because there are some companies that require you as a policy to say that this content is AI developed. So if you don’t, if you say that this is me, first of all, you are lying, of course, but also you are cowardly. Don’t want to say that this is, this is air generated.

1:05:20 – (Joshua Baru): A friend pointed me to Revelation 21, verse 8 which says that, that the Bible says that of the people that are going to be going to hell, the first group will be, will be cowards. So there is sexual immorals, there’s adulterers, there is murderers. But then the first group of people who are going to go to hell are going to be cowards, which was like out of this world then I remember that fear is one of the most. Fear not is one of the most repeated commands in the Bible.

1:05:54 – (Joshua Baru): So it’s, it only warrants why God would send people to hell because they have not obeyed his most repeated command in the Bible. So I don’t want to be, I don’t want to be covered. If I have used AI, I’m going to say I and I want to put myself in that scenario that I have that I’m going to see that I have used there. So I’m not going to, I am going to be very conservative because I’m a researcher. I write a lot, the blogs, the scientific papers. I want to do the actual work.

1:06:25 – (Joshua Baru): So I want, I don’t want to be lazy. I want to work as unto the Lord and I don’t want to be cowardly.

1:06:33 – (Christopher Watkin): The Bible is such an inconvenient book, isn’t it? Like, who would put cowardly in that list? If we were writing that list, we would not put cowardly in there. And yet there it is in black and white in our Bibles. And you’re absolutely right, Fear not is the most repeated command. And what an insightful word that is. Don’t want to be cowards in the way we use air. Thank you so much for sharing that with us. I have a lot to think about.

1:07:02 – (Christopher Watkin): I’d never thought of it in that way before and it’s really powerful the way that you put that. Thank you.

1:07:08 – (Joshua Baru): Thank you.

1:07:10 – (Christopher Watkin): If other people listening to this or watching this, like me, have been profoundly challenged by what Joshua has been saying and you want to take these ideas further, please do come and join us in the Silicon Spiritualities Discord where we’ll be learning together how to follow Jesus in a tech saturated world. You’ll find practical life experiments for every episode and also details of an upcoming online cohort as well, the link’s down there in the show notes.

1:07:40 – (Christopher Watkin): We’d love to see you there. Joshua, it has been a delight to speak with you and learn from you about these things. Thank you so much for sharing your wisdom with us today.

1:07:50 – (Joshua Baru): Thank you for having me.