Part 2: Getting introduced to the invisible apprentice – AI

This post is part of the AI Apprenticeship series:

• Part 1: AI Apprenticeship 2024 @ DiploFoundation
• Part 2: Getting introduced to the invisible apprentice – AI
• Part 2.5: AI reinforcement learning vs human governance
• Part 3: Crafting AI – Building chatbots
• Part 4: Demystifying AI  
• Part 5: Is AI really that simple?
• Part 6: What string theory reveals about AI chat models
• Part 7: ‘Interpretability: From human language to DroidSpeak’
• Part 8: ‘Maths doesn’t hallucinate: Harnessing AI for governance and diplomacy’

By Dr Anita Lamprecht, supported by DiploAI

Week 2 of the AI Apprenticeship online course was all about getting to know AI: its origins, the confusion about its nature, the fields in which it can be employed, and the life-blood of AI – our data.

In confusion, we find clarity

One of the first lectures was about defining AI. It turned out that trying to pinpoint AI in a single definition is impossible and probably increases confusion. AI is a suitcase term encompassing many different levels and dimensions of fields, approaches, and applications, as well as utopian and dystopian visions and ideas. AI is a technology in the broadest sense. It is a technical artifact, a human-made entity that uses natural elements to fulfil a function or purpose.

AI is deeply intertwined with human society and our socio-technological system. Without that system, AI would not exist. To define AI means to define its space in our highly complex socio-technological systems. How can we, as apprentices, handle this confusing situation? By very simply asking, ‘what do we refer to?’ This is a simple method for finding clarity amidst the confusion surrounding AI.

The invisible apprentice

We usually refer to AI as a tool and the subject of study; in our course, we should go further and describe it as an invisible apprentice, like Geppetto’s Pinocchio yearning to understand the world. Why? Because as soon as we start with the practical work, our DiploAI system will learn and evolve with us. This is something we, the human apprentices, need to become familiar with as it is the nature of such AI systems to be more than passive tools.

However, unlike a human apprentice, AI lacks the ability to understand context or meaning in the way humans do. It processes information based on patterns and probabilities, akin to how a flag’s colours and symbols convey specific messages without words.

AI’s everlasting apprenticeship?

AI’s role as an invisible apprentice is evident in various applications, from automating routine tasks to providing insights through data analysis. It assists diplomats by fine-tuning talking points, suggesting alternative formulations, and conducting background research. Will it ever become the master?

For now, it remains an apprentice – capable of mimicking authority but lacking the nuanced understanding and judgment that human diplomats possess. Like Pinocchio, who had to learn and grow through experience, AI needs guidance and feedback to develop its capabilities. So maybe, we should rather speak of an everlasting apprentice than an invisible one. But how to understand AI?

Using flags to explain AI

Explaining the technological functioning of AI is not an easy task. The main lecturer, Jovan Kurbalija, found an innovative way to explain AI through national flags. He showed us how to draw parallels between the patterns on flags and the patterns AI recognises in data. Just as a flag’s design conveys meaning through its arrangement of colours and symbols, AI interprets data by identifying patterns and correlations. This process is central to AI’s functionality, allowing it to make predictions and generate insights.

For instance, consider the ‘colour analysis of flags’ as a way to understand AI’s pattern recognition capabilities. Each colour on a flag can be seen as a data point, and the overall design as a dataset. AI analyses these data points to discern patterns, much like how it processes vast amounts of information to identify trends and make decisions.

Our partnership with AI: Building on metaphors

As we get to know AI, we also need to look at ourselves in this relationship. We often give human qualities to AI – that is, we anthropomorphise it. We compare AI to the human brain or use terms like ‘apprentice’. This isn’t just a linguistic quirk; it shows us how we build relationships. Think about it: we give names to our cars, talk to our plants, and now, we’re developing bonds with AI. It’s human nature to seek connection, even with inanimate objects. The story about Geppetto’s creation of Pinocchio is a reminder of how we try to transfer our own inherent values and ambitions into our creations. These metaphors help us grasp the novelty of AI by comparing it to familiar concepts. They make AI less intimidating and more relatable.

Know the limits

However, it’s crucial to also recognise the limitations of these metaphors. While they highlight similarities, they can also obscure important differences. AI, unlike humans, does not ‘think’ or ‘learn’ in the traditional sense. It processes data based on patterns, without the ability to comprehend context or meaning in the way we do.

By acknowledging these limitations, we can avoid unrealistic expectations and develop a more nuanced understanding of AI’s capabilities and limitations. It’s like any relationship: it requires clear communication.

Conclusion of Week 2

As we progress towards the practical part of our apprenticeship, understanding AI’s role becomes increasingly important. By using metaphors like flags, and remembering the lessons of Geppetto and Pinocchio, we can demystify AI and make its complex processes more accessible. This understanding is essential for informed, inclusive, and impactful debates on AI governance as we strive to anchor AI in the core values of humanity.

The AI Apprenticeship online course is part of the Diplo AI Campus programme.