EP Co-Worker Retreat - Workshops

WORKSHOP 1 (Live demonstration of AI)

For the first workshop, the speaker of camp, Mr Stephen Driscoll did something different - a live demonstration of AI!

He began by doing a broad survey of the attendees - asking them to rate their competence with AI from a scale of 1 to 10. He then got those with the highest rating to share what they have done with AI. The examples given were very encouraging - one pastor even used AI to come up with an app for his church! 

Stephen stated that there are different levels of AI - not all AI is the same:

• Level 0 is the Google AI assistant which comes with Google search  

• Level 1 is free conversational AI, such as Chat.GPT, Gemini and Claudes

• Level 2 is paid AI (Stephen uses Chat.GPT) which remembers one’s previous inputs. 

• Level 3 is using paid AI for functions like codex, cowork, deep research 

During the live demonstration, he gave each level of AI various questions, such as ‘What are some examples of anecdotes that appeal to people from Singapore?’, and ‘In Ephesians 1:1,  how many historic commentators really saw "saints" and "faithful" as 2 audiences?

From difference in the quality of the responses, we saw first hand how the various levels of AI makes a huge difference.

Stephen then showed some projects he has worked on using AI. For example, he used AI for sorting out rooming and scheduling arrangements for a camp, which he would previously spend days working on. Also, he gave AI (through codex) access to his database of talks and sermons, and AI will flag up if he uses an illustration or example he has used before in that same place. This is because he speaks in many different places, its hard to take note if he has repeated himself to that specific audience before.

Stephen addressed the question as to whether pastors using AI for sermon feedback is helpful or harmful. He notes that LLM (large language models) like Chat.GPT or Claude, as the term suggests, are extremely competent in language - its like consulting an expert in rhetoric or pedagogy. 

The problem is when we start handing over theological control to AI. He notes that AI is trained to be inoffensive - and it will often suggest cutting away the most controversial point from a message, even though that’s what the congregation may need to hear. 

Finally, Stephen concluded with some thoughts, after a very convincing demonstration of the power and flexibility of AI to us all. They are: 

1) long term efficiency means short-term inefficiency (to pick up AI, and involve it in our tasks, we need to 'waste' time on it)

2) trying and failing is better than not trying 

(some people have a fear of picking up a new skill, but actually AI is relatively simple to learn)

3) data and privacy 

If we use AI for ministry work, we must be careful about allowing it unfiltered access to sensitive data

Stephen added that many pastors and ministry staff face heavy workloads - and many will be greatly aided by AI if we let AI handle the more adminstrative tasks, freeing us up for more pastoral and relational tasks.

WORKSHOP 2 (Technical & Practical aspects of AI)

For Workshop 2, held on the second night of the retreat, Stephen covered the technical & practical aspects of using AI. 

He noted that whilst some pastors may be adverse to, or not interested in using AI in relationship to Christian ministry, those whom we lead - ministry staff, lay leaders and members - may not have such qualms, and have to be educated on its proper usage. 

Firstly, Stephen addressed the defeaters - what common arguments people might have against the usage of AI, with 4 observations:

1. Ministry has always occurred in the context of change

(An example is the printing press, which helped to accelerate the Reformation)

2. There is nothing in principle wrong with technology 

3. Technology does not in principle replace and threaten the role of the Holy Spirit 

(The Holy Spirit can work through various technology - culture, media, radio, powerpoint, and is not diametrically opposed to it)

4. There's nothing in principle that's wrong with efficiency 

As such, we should not shy away from AI as a technology and common grace.

Secondly, Stephen noted the dangers that AI posed - especially in the ways it affects the thinking and reputations of pastors.

He proposed the scenario: By 2028, if AI is able to write sermons that have better introductions, are more exegetically sound, and are more humorous than pastors, will that mean we should leave sermon-preparation entirely to AI?

Stephen brought us back to Deuteronomy 6:4-9, which mentions that knowledge of God’s word has to be both external (write them on doorframes … ) and internal (are to be on your hearts …). An MIT study has shown that students who lean on ChatGPT struggle to remember quotes or even the main points from their essay - there is a difference between the thinking that arises from the hard work of writing/speaking, and copying and pasting a final output. As such, we must be wary that AI does not replace our thinking. 

As for AI doing damage to our reputations, he gave the scenario of pastors sending consoling text messages to their members using AI, or depending on AI entirely for sermon preparation. How would congregants feel if they found this out? Intuitively, there will be a sense of betrayal - and that the pastor is purely performing and even plagiarising.  

Where then should the line be drawn for plagiarism from AI? Stephen contends that it should be guided by 3 Ps:  

1) proportionate - how much of the sermon was done by AI

2) purposeful - how intentional was the reproduction of AI content

3) personal - i.e. anecdotes about speaker's kids should have happened to the speaker's actual kids, not be an invention by AI

As an antidote to the temptation to use AI wrongly, Stephen proposed some practical handles:

a) Preachers should ‘chew-first’ - prayerfully read the Bible and commentaries, and form independent thoughts before leveraging on AI.

b) Preachers should take ‘slow-steps’ - even though AI may offer instant answers, we should meditate on God’s word, and prayerfully consider the needs/circumstances of our people. 

c) Preachers should endeavour to re-express. Even if AI may produce smoother statements, he prefers that preachers use their own words in their ‘voice’, even though it may be 5% clumsier.

d) preachers should ‘argue back’, and probe whatever answers AI gives. He states that AI are meant to be conversation partners, not oracles. 

For his final major point, Stephen talked about ministering to an AI riddled world, even as AI use proliferates. 

Firstly, he notes the individualization and personalization of entertainment, news and religion.

In the past, families would watch the same TV serials, but now, everyone can choose their diet on their screens. This erodes the common ground and shared spaces of individuals, who are now more inclined to look for communities that fit their needs exactly.

Secondly, Stephen showed a graph on how social media rewards extreme views with more hits, because that catches people’s attention. This leads to polarization - where people are unable to disagree. In constrast, scripture counsels "Faithful are the wounds of a friend …"  (Proverbs 27:6) - disagreement and friction are hard but good things 

Finally, Stephen pointed out reverse polarization, which social media also proliferates. Occasionally, the algorithm will feed a right wing person (for example) the most extreme left wing views. This only increases distrust and widens the divide between groups.

He points to research showing that the gap between genders (viewpoints between males and females from the same age range) is widening in every country, and the AI algorithm is a key driver in this. He says that what’s on our feed (what we consume on social media) can offer a mirror to the depth of discipleship.

Overall, through the workshops, Stephen gave very insightful handles on the opportunities and threats that AI affords. May God give us wisdom as we harness and direct this technology, for His glory