Artificial Intelligence
Artificial Intelligence
23rd July 2023
Mark 13:1-8
I’ve noticed two reactions to Artificial Intelligence (AI):
1) enthusiasm – students who can get ChatGPT to write their essays, curiosity about what AI can do.
2) fear/apprehension – people who have seen the dystopian Hollywood movies of an apocalypse brought about by sentient robots rising up against humanity
These responses kinda match the earlier responses we’ve seen to social media harvesting one’s personal data:
· some enjoy the connections social media offers and suggestions of products (that suit your personality and waist measurements!), and
· some are alarmed at the ability of the technology.
One thing that concerns some, is that the abilities of AI are unknown.
In recent months we have been made aware of the rapidly expanding presence of AI,
and the predications some are making about the influence AI might have in our world.
One prediction about AI recently grabbed my attention:
“AI can create new ideas. [It] can even write a new Bible.”
Yuval Harari, an Israeli author and academic says we can legitimately expect from AI “super-human intelligence”.
Now, I know a bit about the Bible (you’d expect), but we have someone else who knows a lot more than me about AI…
I want to ask Neil Dodgson to come and answer some questions.
Neil is Professor of Computer Graphics at Victoria University, and from many conversations I’ve had with Neil, I really appreciate how much he understands about what AI is (and isn’t).
a. What do you make of this comment about a super-human intelligence writing a new Bible?
It is wrong on three levels. First, he is assuming that we can build an artificial super-human intelligence. We’ve not done this yet and the jury is out on whether we could ever do it. But let’s assume that there is some future super-human artificial intelligence. He’s then assuming that such an intelligence would be able to write something that was the definitive truth. Humans have been searching and debating truth for thousands of years. What gives me hope here is that the debate about what it means to be human and the meaning of life has been going on for so long and yet things written before Jesus was born are still useful and debated. Maybe there is no magic answer that we could find if only we were intelligent enough. And the third level of criticising is that he’s creating this imaginary machine in his own image: he’s assuming that the machine would write a text that was true. Past experience is that it’s much more likely that the machine would write a text that allowed it to control human beings.
b. What is AI anyway?
Artificial intelligence is any computer system that attempts to replicate what we call intelligence. So, playing chess, understanding human speech, translating between languages, recognising faces, or writing text. These are all examples of intelligent behaviour and we have built computer systems that can do all of them. Old fashioned AI was done by hand programming computers to reason in the way that we thought humans reasoned: that’s how we built the first chess-playing computers. Over the last 20 years we have transitioned to neural networks. Those are inspired by the structure of the human brain, where millions of individual neurons are connected to each other by billions of connections. Each neuron is simple in itself. But the combination of all of them together creates what we call an emergent behaviour, where the overall behaviour is more than the sum of its parts. Today we build massive neural networks and then get them to do specific tasks by feeding them enormous sets of training data. Each piece of training data tweaks the network a little bit so that, after a lot of training, the network is tuned to do the task you want. For example, you can train a neural network to recognise photos of animals by training it on millions of photos of different animals and telling it what the answer should be in each case. This is a photo of a dog, cat, goat, etc. At first the network just guesses randomly, but as you train it it gradually gets better. Once trained you can show it a photo it has never seen and it will have a very good chance of getting it right.
c. I’ve heard of ChatGPT, what does that do and how does it work?
ChatGPT has been in the news a lot over the last nine months. It is a neural network system that can produce beautifully written text. Ask it for 200 words on New Zealand politics or a poem in the style of William Wordsworth and it can do it in a few seconds. It is trained on literally billions of examples of human-written text including tens of thousands of freely-available books. It produces really convincing text but all it really does inside is string one word after another. What it has been trained to do is understand which words generally follow other words. It keeps track of the previous several thousand words and can thus generate plausible sounding text based on that context. The problem is that it has no understanding or knowledge behind its utterances. It truly does just string words together. I’ve seen it produce amazing material that we would call very creative if a human did it, and I’ve seen it produce the most ridiculous nonsense, such as telling me that Peter Snell was the first person to successfully cross the Cook Strait on foot — something that is actually impossible — but giving me complete beautifully-written paragraphs explaining this impossible feat.
d. Are systems like this going to lead to a super-intelligent computer? And should we be worried?
Well, from how I’ve just described ChatGPT, you would think that the answer is no. How can something that just strings words together become intelligent. However, as I’ve said, the jury is out on this one. Tests with ChatGPT, and more powerful language models, show that these systems are exhibiting behaviour that looks intelligent in many ways. While they are just stringing words together, if you give them the right context, they can produce output that is good or very good. These tools have passed tests that many humans find hard. That is why the jury is out. From this evidence, some experts say that what we are seeing is an emergent behaviour: intelligence that arises out of an incredibly complex system doing something that seems to be more than the sum of its parts. They say that these systems are doing today, at a simple level, what human brains do. Once we make a system of sufficient complexity, they say it will have intelligence indistinguishable from a human brain. And then will go beyond that to super-intelligence. But there are others that say that these systems are a dead end in developing truly general artificial intelligence. That these systems are just stringing words together and that this is not the way to get to a super-intelligent computer. So can we build a super-intelligent computer? We just don’t know.
Should we be worried? Yes. We may never have super-intelligent computers but even the existing AI systems have the potential to profoundly change society. In the nineteenth century steam engines changed how manufacturing was done. In the twentieth century mechanical diggers changed how many people you needed to build a road or railway. In this century, these AI-based tools change how we use our brains and what jobs are needed. Already we have AI systems doing medical diagnosis, deciding whether you get a mortgage, and recommending what to watch next on YouTube. We do need to be concerned about how these tools are used and how much authority we give them to make decisions about our lives.
You’ve given us a succinct summary of what this technology is, …but we also need to think about how we choose to relate to it.
e. What do you think we, as Christians, should do to engage with these developments?
We could do two things. We could go down a theological route of thinking about the implications of how we would handle conscious, thinking machines. There are connections here to the church’s historic debates about how we treat animals, with how we handle slavery, and with the nature of humanity. That would be interesting but I think the second route is more useful. The second route is to think about how AI systems affect society. This is again something where the church has a lot to say. We have a solid foundation on which to rest our arguments: we fundamentally believe in the intrinsic worth and value of every human being. We can discuss and debate how will these AI systems change how society works, how humans are treated, how the country is run, and how human beings treat one another. On what basis and for what purpose do we use AI systems? Fundamentally, how much are we going to abdicate responsibility to machines? How much authority do we offload to our computer systems?
You’ve mentioned how much authority we choose to give this technology.
And this is an important feature of what it means to be human: what do we consider authoritative?
What do we trust to guide us in our decision-making?
AI (at least some of the claims about AI) is a contemporary example of this ancient issue of: ‘what do we consider authoritative?’
Will AI technology be something we choose to give authority?
How will we allow it to guide the way we live?
Make no mistake, regardless of what AI technology is actually capable of, it is a choice we make to make something authoritative in our lives.
This has always been a crucial feature of how we understand and express our human autonomy and freedom.
Consider the shift of authority that occurred through the Protestant Reformation.
Basically, the Reformation relocated primary authority in the Word of God.
This belief continues for our Presbyterian Church.
In our contemporary confession of faith, The Kupo Whakapono) we declare that the Scriptures are “authoritative for our faith and life and are our supreme standard of belief and practice.”
We trust the Bible as the best authority for knowing and responding to God’s will for humanity.
One implication of this is that the Church isn’t about pre-packaged answers.
As we are attentive to the Word of God, with the guidance of the Holy Spirit, we navigate questions about culture, about justice, about our purpose on earth.
This is both part of the human pursuit of truth, …and also will sometimes conflict with other loyalties.
In today’s Bible reading, Jesus warns that all that appears great may not be great for ever.
The buildings (that the disciples are so dazzled by) won’t last.
Jesus warns them:
“Beware that no one leads you astray.” (v5)
Do we hear this warning for us, for what we make authoritative in our lives?
Jesus says
“Seek first the Kingdom of God and his righteousness, and all these things will be given to you as well.” (Matthew 6:33)
When we mistakenly seek after and trust something other else as authoritative, the Bible warns us this is… idolatry.
Idolatry is when you make an idol by elevating anything to be more important to you than God.
A common mistake people make when they hear about the biblical concept of idolatry is thinking that idols are bad things. But that is almost never the case.
The greater the good, the more likely we are to expect that it can satisfy our deepest needs and hopes.
So, we encourage one another to recognise that we find all we need in God.
God is our appropriate authority – every day of our lives.
Our trust in God will be misunderstood and even mocked, but this loyalty gives us our identity as participants with God in the reality Jesus describes as ‘The Kingdom of God’.
This trust in God compels us to strive to be a community reflecting nothing less than the glory of God.
Our St John’s Vision includes creating ‘safe spaces’, to foster the reality of the Kingdom of God.
How do we do this?
Well, our primary responsibility as the Church is to be a people formed through shared practices of worship – shaped by the story and the sacraments.
Gathering together, learning and repeating the practices of faith solidifies our loyalty, enabling and equipping us to actively and creatively bear witness to God, and God’s vision for humanity.
When we recognise God’s authority and God’s vision for humanity (far from narrowing our view) our horizons broaden.
The more we trust God’s authority, the more we can anticipate the reality of the Kingdom of God emerging all around us; trusting God’s vision for humanity which will be one day be fully realised.
What does this mean for us every day?
Let me conclude on a very practical note.
How can your actions match the recognition of God’s authority?
· You can evaluate the claims about technology (like AI) against your loyalties.
· You can check your assumptions about what is good and worthy are consistent with your most deeply-held convictions.
· You can make decisions that give witness to your willingness to recognise God’s authority.
· …You can exercise your vote thoughtfully at the General Election.
Voting is one expression in our society of how we recognise God’s authority, if we choose to carefully examine kingdom considerations.
And for this, along with any decisions we make as Christians, we have rich resources of faith available to us.
This isn’t about pre-packaged answers.
It’s an expression of freedom …and imagination …and responsibility …and purposefulness.
To come back to the main message today:
authentic decisions are grounded in what you consider authoritative.
David Gushee suggests:
1. (First and foremost) consideration of scripture as read through the (prophetic) lens of Jesus Christ and in the manner of Jesus Christ.
2. tradition
3. direct divine guidance
4. church leaders
5. reason and science
6. experience
AI and other technologies will remain part of our human experience, and it’s up to us to relate to them in a way that recognises modest authority, as we seek first the Kingdom of God and His righteousness.
We join with countless billions who declare boldly: “Jesus is Lord!”