AI systems are gaining ubiquity in certain parts of life. This isn’t always as useful as it seems and could be doing long-term damage to our decision-making ability. Is AI creating ignorance?
An everyday example
Do you use a satnav when driving? Satnav gives the promise and usually the experience of fault-free guidance to a destination when it’s an unfamiliar one. The geo-spatial technology behind it is very successful at doing so.
Advances in real-time tracking mean we’re no longer saddled with first-generation discrete satnavs; intelligence in the cloud allows the benefits of traffic re-routing and dynamic additions such as road works. This is now the norm while we still manually drive.
But is there now a reliance on generally out-sourcing this type of problem-solving?
Pitfalls to consider
In user-centered design (UCD) the successful use of satnav is literally what’s known as a ‘happy path’. We arrive at a predicted time and location, and the tool is almost invisible. Reliance on this makes it progressively harder to revert to manual navigation. But when exceptions to the ‘happy path’ are encountered, cognitive dissonance can be easily triggered and mistakes made.
Take the non-sensical, humourous (or sobering) story of a man that actually followed his satnav guidance into a river.
What’s going on here is more than a lapse of driver judgment or attention, it’s a deeper problem at play.
AI directs us so well that it masks potentially significant decreases in cognitive ability.
We have learned to trust AI
The beauty of AI is that it offers unprecedented utility functions to huge swathes of the global population. We have learned to trust these systems, sometimes overriding our direct perception. Seeing flowing water but still driving into it sounds counter-intuitive, but if you’ve ever been in a fast-decision situation with a confident voice telling you what to do, ignoring or disobeying isn’t as simple as it sounds.
This isn’t submission to authority per se, such as the infamous Milgram electrocution test, but a more nuanced issue of trusting technology as a cognitive shorthand. The smoother the experience, the less we question the system, and the more agency we lose. When it goes wrong, choosing a different response is mentally harder to do than complying. This has implications far beyond car navigation.
Becoming more cybernetic
When AI gives us the answers, we are not learning as individuals. Through passive consumption of AI suggestions, we are becoming ever more cybernetic, symbiotic – technology is the structure we grow around. AI also helps us decide what products to buy in daily life, removing yet another layer of agency.
Is there anything wrong with this? Should we be concerned that AI is creating ignoance?
Acquiring knowledge gives autonomy
A famous Chinese proverb:
The meaning is very simple – acquiring knowledge and applying it practically gives autonomy.
When we teach, the original human way, it’s an interactive modelling process of ‘show and tell’. Trial and error. At its historical evolutionary best, it’s a connective, social bonding process.
Teaching and learning is a dynamic activity – a teacher initiating activity by passing it into another person, who practices replicating the teacher’s ability. Mimetically transferring ideas and thoughts from one mind to another. This is qualitatively different from a YouTube walk-through.
The students are not passive recipients. In this example, they learn how to fish. It’s an intrinsically active, interactive experience. This has a survival benefit. It’s a socio-emotional algorithm; they can adapt ways of fishing because it’s a technique. As this is a process it requires effort. It’s forging new neural pathways, increasing the cognitive complexity and capability of the student.
AI helps us avoid effort
This effort is the very thing AI helps us avoid. In fact, AI tries to help us in a different way – more like “Give a person a fish every day, they never go hungry”. You don’t need to teach them anything. They don’t need to learn anything. The consequence of this approach is that they never learn to fish.
People could spend that time on other things such as solving bigger world problems.
But, more commonly, we see a large-scale increase in activities like time spent on scrolling social media, and associated rise societal mental health issues, particularly for younger generations.
Let’s turn our attention to Google
It gives instant answers as declarative information and procedural ‘how to’ broadcasts.
The individual interprets ‘teaching’ at a distance without the critical feedback loop of direct interaction. We, as information seekers, become dependent on being ‘given to’ by the algorithms, on-demand. This utility comes at a cost because neural pathways grow through active learning experiences.
A well-known example of this would be London taxi drivers who did ‘The Knowledge’, meticulously learning by practice the intricacies of London roads. This gave them improved memory function, evidenced not only in ability but physiologically. The hippocampus grew.
Nowadays they use satnavs instead. Local, internal knowledge has been replaced by dependence on external supply. Here the phrase ‘use it or lose it’ comes into sharp focus.
Without practice, the neural pathways needed for problem-solving and critical thinking degrade over time. This is also physiological – atrophy of the myelin sheath that facilitates neural connections means certain pathways don’t get reinforced and particular kinds of thinking become more challenging.
Information on demand
Google gives information on demand. Satnavs give directions on demand. Here’s the crux of why this is a problem;
When I lived in Japan and mobile phones were first becoming popular, a common lament by the older generation was that children were losing the ability to write the language. They could select from suggested characters in text messages, but skill with pen and paper to write them out was fading.
This is anecdotal but as I expected is now being borne out in scientific research which found that “writing by hand increases brain activity in recall tasks over taking notes on a tablet or smartphone. Additionally, those who write by hand on paper are 25% quicker at note-taking tasks than those who use digital technology”.
AI gives us time and we passively consume
AI suggestions and support remove our need to activate certain brain pathways. What is the ostensible benefit to us? Time. What do we do with the ‘free time’ AI gives us?
Without the foundational pillars of critical thought, mediated by embodied learning through experience, our brains effectively default to the lowest energy use through passive consumption.
Practice makes expertise
Practice makes expertise and by definition, it’s not an easy process. But it is rewarding. I’m not suggesting everything we do needs to be first-hand experience, such as navigating by stars, but combining the passive process of consuming content with real-life activities, real-life engagement, gives more context – and context means wider perspective, which means a greater surface area for understanding and creativity. This is the well-spring of progress if not success.
Is AI creating ignorance?
AI gives us automation and answers, moving into areas previously only done by qualified humans such as determining government policy, housing decisions, and the criminal justice system.
But over-reliance on this will erode the functional neural networks that give us our unique ability to process information, and vitally, to formulate both meaningful questions and novel interpretations.
Let’s avoid Artificial Ignorance.
I’d be happy to hear your thoughts on this blog, please drop a comment below. Thanks, Paul.