What is a lie?

I always thought that a lie is a bold and shameless falsehood. It turns out that a lie is just a complement of truth. And given the high bar of truth, anything may contain lies. That’s in fact the title of Alex Edmans’ book: May Contain Lies.

Lies are everywhere; they come wrapped in facts, backed by data, supported by studies, or amplified by authoritative voices. And they feel true, because they flatter what we already believe.

I spent my teenage years devouring behavioural economics books. Daniel Kahnemann, Dan Ariely, Richard Thaler, V Raghunathan. This resulted in a vast zoo of behavioural biases that I always tried to keep in mind when thinking things through. That was rather unwieldy, because it is hard to go through an extensive checklist of biases for every thought or decision.

In May Contain Lies, Alex Edmans argues that when one already has a point of view on something, then Confirmation Bias is a huge danger, and otherwise, when one doesn’t have a point of view, then Black and White Thinking is lurking around.

Confirmation Bias is a cognitive bias where people tend to search for, interpret, favor, and recall information in a way that confirms their pre-existing beliefs or assumptions, while giving less attention to or dismissing contradictory evidence.

Black-and-White Thinking, also called all-or-nothing thinking or dichotomous thinking, is a cognitive bias where people view situations, people, or events in extreme, either/or terms, ignoring any middle ground or nuance.

Alex Edmans argues that by being aware of just these 2 biases, one can improve critical thinking across the board. That sort of parsimony is rare and enlightening — and boy, do I love the Pareto principle!

Reality is not a flat surface where our thoughts wander. The geometry of thought has some steep slopes shaped by Confirmation Bias and Black and White Thinking. Being aware of this nature of our minds makes the self-correcting nature of critical thinking essential for not slipping into a vortex of lies.

A useful framework Alex Edmans introduces is The Ladder of Misinference:

* A statement is not fact: it may not be accurate.

* A fact is not data: it may not be representative

* Data is not evidence: it may not be conclusive

* Evidence is not proof: it may not be universal

What May Contain Lies by Alex Edmans taught me more than anything else is that truth-seeking is a mindset, a process, and a way of life.

And given how easy it has become to generate text/multimedia, the price of creating misinformation is tending towards zero. And with that, the need and ability to think critically is becoming increasingly priceless.

Watch the full video episode:

Follow Alex Edmans on LinkedIn: https://www.linkedin.com/in/aedmans/

Find out more about May Contain Lies: https://maycontainlies.com

Buy May Contain Lies by Alex Edmans: https://www.amazon.com/May-Contain-Lies-Statistics-Biases_And/dp/0520405854

How I Apply This to “AI“ Right Now

To bring greater clarity to the current “AI“ landscape, here’s a summary of how I applied what I learnt.

1. Uncover vested interests

Anyone selling “AI” or to “AI” is probably bullish due to their incentive to sell.

Any investor in “AI” companies that has deployed lots of capital and needs growth.

2. Ignore extreme sensational positions that ignore nuance

Hypists: The world has changed dramatically because “AGI“ has been achieved.

Doomers: the world will end because of “AI“ or “AGI“.

3. Be honest about confirmation bias and its origins

I am trained as a mathematician with a research specialisation in machine learning. Over a decade, I have operated by training and deploying learning algorithms and am thus sceptical that just scaling compute and data will result in “AGI“

4. List all nuanced points of view and explore them with curiosity and deliberate open-mindedness

I make an effort to gather a wide range of perspectives, especially those that challenge my views. I approach them with curiosity, not defensiveness, and remind myself that no one has a monopoly on truth.

5. Articulate the smallest set of perspectives that are exhaustive and as orthogonal as possible

I try to reduce the noise by mapping out a minimal set of distinct and independent perspectives, those that together cover the landscape without too much overlap. This avoids drowning in details and helps sharpen thinking. Something similar to a Principal Component Analysis in mathematics.

6. Reflect and do thought experiments to simulate in my mind the nuanced implications of each point of view

I deliberately imagine how the world might look in each scenario. I ask myself: What changes about how value is created? What changes about how value is captured?

This mental simulation forces me to think through second-order effects, not just headlines.

7. Try to identify gaps in knowledge or insight

After going through this process, I focus on spotting my blind spots.

Where am I the most certain or uncertain?

What areas feel underexplored?

Where do experts seem to disagree sharply, or where am I relying on assumptions that I haven't fully tested?

Next on Part-Maven Part-Maverick, we will discuss with Sangeet Paul Choudary how value is created in the age of AI, who captures it, and what the implications are for businesses and us as humans.

This Substack is reader-supported. To receive new posts and support my work, consider becoming a free or paid subscriber.

Subscribe to be first to know when the episode drops: https://www.youtube.com/@SLASOG

For more of my thoughts, follow me on LinkedIn: https://www.linkedin.com/in/rritavan/

Get my book Data Impact for a pragmatic take on data-driven value creation for business: https://www.amazon.com/Data-Impact-businesses-LEVERAGE-SIMPLIFY/dp/178133921X/

Get full access to Part-Maven Part-Maverick at mavenmaverick.substack.com/subscribe