The Need For AI Literacy – Two Analogies

There is no need to say that AI is all around us. ChatGPT has brought AI to the forefront of just about everyone in the past few months. The potential for disruption, value, and harm, is difficult to overstate. As everyone from technologists to politicians to philosophers weigh in on the challenges of the moment, it may be worthwhile to consider other technologies that have transformed our lives – some from afar and some from very close up. What can we learn from them? In this article, I cover two analogies, automotive technology, and nuclear technology.

Why automobiles?

The figure below is that of a 1909 Vauxhall B-type 16-hp semi-racer. To drive this vehicle, one needed to understand how to regulate the air-fuel mixture, pump the brakes by hand, and so on. If you did not understand how the engine worked and how to manipulate it – the car would not run. Fast forward to where we are now (even before self-driving cars). Most of us could not explain how a car engine works, but every person’s life is impacted every day by a car. Even if you are not driving one, you may be taking a bus, your food may be delivered to your store by a truck, an ambulance may take you to a hospital, and the list goes on.

Cars in 1909 are not too different from AI till very recently. Only those with some understanding of the internals could get AI to behave as they would like. With technologies like ChatGPT, however, AI is moving rapidly into the mainstream, affecting our lives whether we understand how it works or not.

So – what can we learn from cars? Literacy is key.

I believe what we can learn is that society has developed a level of literacy about automobiles that enables us all to function relatively productively in the presence of a variety of automotive technologies.

• We know how to select what we want, evaluate our options, and make build vs. buy decisions
• We can assess the level of education we need (do we need to learn how to drive a car, a truck, a motorcycle), and how to get the education we need. There are ways to educate young people and bring them into the fold.
• There are laws and mechanisms to check our understanding of how to safely interact with technology. The laws match the risk level (seatbelts for riders, driver’s licenses for drivers, etc.). There are mechanisms in place to assess the risk of the devices themselves, and keep the broader community aware of changes in these risks.

AI Literacy is the equivalent for AI. I expect that AI Literacy will be a requirement for every human – to help us navigate the daily interactions we will all have with AI.

Nuclear Technology

Why nuclear power? While cars are a useful analogy for technologies that affect us close up, nuclear power is a better analogy for a technology that is far away from most everyday people but has the potential for massive (positive and negative) impact. If we look at the history of nuclear technology, we see everything from the positive intentional impacts of powering hundreds of millions of homes, to the accidental negative impacts of disasters like Chernobyl, to the intentional impacts of nuclear weapons – which are positive or negative depending on who you ask. We also see that managing the risk is not a one-shot deal. A cursory skim of recent world news tells us that even after many decades, managing nuclear risk is still an everyday global challenge.

What can we learn from this? Also, in my view, literacy – but perhaps of a different form.

AI has every bit of positive and negative potential as nuclear technology (and possibly even more so), in areas most of us do not deal with every day. For example, AI has already shown the potential to generate new drugs faster than any process before it. It can also generate new chemical weapons faster than any process before it. It has geopolitical implications as nations race to get an edge on AI technology in every aspect. One can expect that the struggle to manage AI risk will be as complex and arduous as the challenge to manage nuclear risk continues to be.

It also speaks to a different kind of AI literacy. The cars example showed what everyday literacy can need to look like from a consumer perspective. There is also an underlying literacy concentrated on manufacturers, government oversight bodies etc. Nuclear technology takes this policy aspect further and demonstrates a level of literacy, education, and policy that has to cross multiple levels within a country and across countries.

Takeaways

It is quite possible that the AI challenge is greater than any we have encountered before. Automobile control is accessible to the average human but a single car cannot destroy a planet. Nuclear technology is still not accessible to the average human. AI has the accessibility of the first and the reach of the second. AI literacy and AI Education will be required for humans to be able to voice their opinions on how AI should integrate with humanity, with a basic understanding that will help them know and express what they would like to see for themselves and their children.