The allure of rabbit holes continues. And when I say “rabbit holes” what I really mean is AI seducing me into a veritable rabbit warren of possibility. Endlessly. A simple question forms and reforms into an endless chain of related questions which feeds my innate curiosity which, some days more than others, is hard to curb. “Oh, I’d never thought of that angle” and so on.
Just like social media the system is designed to keep us sucked in, to explore finer and finer aspects of something that was never designed to be analysed in such detail. It has all the hallmarks of being stickier than glue. It responds instantly with unlimited and personalised follow-up questions, suggesting new topics and connections with detailed explanations, giving examples and comparisons on demand. There are no natural stopping points like the end of an article or chapter. It unfolds relentlessly on.
I’m curious about the addictive nature of the process together with the effect of intermittent reinforcement. When we discover something new or find the exact words for an idea, together with all the relevant references complete with links, how can we resist? That aha-moment of a confusing subject becoming clear, or discovering an unexpected connection can be akin to an adrenaline hit. But the cost is mighty. We may be tempted to delegate deep thinking and decision making to a machine. There’s little opportunity to pause and reflect, take a breath, stretch, walk outside, get it all in perspective, it just keeps sucking us in.
Intermittent reinforcement contributes to the process – when rewards arrive unexpectedly rather than every time, we stay for just one more answer or another, then another, just in case a “brilliant” answer emerges from the soup of everything everyone knows through all time. Well, I’m exaggerating, but AI entices us into that perception. It presents its answers with such utter conviction. Yes, it hallucinates from time to time or dishes up slop, but even that is occasionally amusing.
I’ve come to the conclusion (incidentally, without its help) that the key concern is less to do with whether the information is useful, it is whether the ‘research’ (ha, ha) continues to serve its original purpose? Am I simply collecting more and more mountains of information without making any progress toward whatever goal I had at the outset? Wait, what was that goal?
So I decided to ask AI how to create an exit and I paraphrase its pretty good strategies:
- Clear definition of what I want to produce at the outset
- Setting a time limit, or a scope limit
- Instruct AI to summarise rather than keep introducing new fascinating topics
- Take note of some of the interesting tangents and review them again tomorrow for relevance (a 24-hours later perspective is revealing)
- Be clear and strategic about asking myself the question “is this enough?” Is now the right time to make my exit?
- Can I allow nature to beckon with its simplicity and history and beauty?
That’s probably the most important question right there, in fact the one I came to at the conclusion of Rabbit Holes (Part One) last month. How Ironic.
[My private practice is generally full but you’re welcome to reach out for an update on my current availability].