Here we are again. It is once again time for the next installment of the artificial intelligence sales pitch; the technology that will undoubtedly disrupt the world as we know it today five years from now, just like it was going to do five years in the future from last year (and the year before and the year before that…) . I bet you heard this story many times over by now, and I understand it’s tiring. I can appreciate how some people are tired of sitting around with their digital muskets asking, “So when exactly is this AI revolution supposed to happen?” Can we blame them? As it stands, our brains naturally gravitate to great and compelling stories, and frankly, artificial intelligence is one that has done a tremendous job at captivating our minds (both in terms of near-utopian potential and dystopian nightmare scenarios). I must admit that I wasn’t completely able to avoid drinking the Kool-Aid either. When I stumbled across a company that made use of AI in the mineral discovery process, my imagination began to take hold of me. The AI fantasy (or better termed hallucination… that’s how unrealistic it was) could be summed up like this: “Artificial Intelligence would take all existing data inputs and the algorithms would then pinpoint precisely where the world’s biggest gold deposits would be located.” Voila! Mineral exploration so cheap and easy it could be done with the click of a button and doesn’t even require getting out of your pajamas. That was the story, and if true, talk about a sexy story! Unfortunately, while it might be as sexy as a party at the Playboy Mansion, it was also just as fake…

While AI can’t live up to the utopian delusions, that doesn’t mean its use and value should be completely discounted. AI’s use in exploration has been integral in producing some incredible results along the way. The important task is resolving the big discrepancy between many people’s expectations and real-life results. If left unresolved, the investor delusion begins to set in as we are left to question, as I did myself; how can a company using the most cutting-edge advanced AI, NOT consistently make world-class discoveries, like, every month?

It really is one of the flaws of the human brain. Our primitive brain likes, no, LOVES appealing stories. We are guided by stories - while we would like to think we are built on logic, the reality is that we are actually built on narratives. Religions, politics, you name it - they are all stories. And I don’t mean that in a belittling way - it is simply a reality. The stories are all around us, and frankly we are helplessly addicted to them. Not surprisingly, AI is a story we really like. I mean, just look at the concept: A self-improving, learning and adapting system that can handle practically any theoretical problem you feed it. How does that not speak sweet nothings to our hearts and minds? While conceptually pleasing, it is driven by a tremendous fallacy. I was not aware of how big this fallacy actually was until I experienced it firsthand.

After having my AI fantasy bubble burst, I was left dissecting how I had allowed myself to become so enamored with it in the first place. Well, it turned out that I had - again - been tempted by the sexy narrative. AI is an exciting buzzword nowadays. Every company seems to be developing its own proprietary AI systems, and touts how it is seeking to solve the world’s biggest problems using its AI solutions. From Full Self Driving to Precision Agriculture, AI is bantered around as the miracle cure for every major problem. And while eventually, some of these AI solutions may one day become a reality, the narrative that surrounds AI these days can certainly leave someone with the false impression that those times are already upon us.

And frankly, who can blame the person who falls for the narrative? The percentage of people who have a proper understanding of how AI actually works is actually quite small. Ultimately, not everyone can be a computer or data engineer. Companies know this too - and some are willing to exploit it. The companies have to sell themselves in order to survive. And what better way is there to catch someone's attention than by speaking to their imagination? That is much more appealing than just laying out the cold hard facts on artificial intelligence; namely that AI is just a bundle of mathematical techniques. Most people associate math with boredom. Just think back to your math class in high-school. Who is excited by that thought? And besides, math doesn’t sound revolutionary, or innovative - actually quite the opposite. Frankly, that’s precisely the first thought that entered my head when I heard an executive from an AI company say these exact words: “AI is not some kind of wizardry, it’s just math”. I was shocked. Deep down, I knew it was the truth, but the urge to resist that truth was surprisingly strong. It sounded so old-fashioned and downright unappealing. How dare this company sell themselves as a math company? What is their next big project? An updated edition of a high school textbook?!?! This admission made the company sound like the losers of the pack. Other companies were creating innovative solutions that were going to magically solve the world’s problems… and this company was busy fine tuning its math equations? Ugh!

It was certainly not an instant (or pleasant) epiphany for me, but over time as I thought about this statement, something began to sink in. Maybe this company was not the loser. Maybe the rest of the pack were the hopeless dreamers who were bound to overshoot the mark and leave me holding the bag of their starry-eyed ambitions. It made me reflect back to a famous speech given by Peter Lynch to the National Press Club. Here is the excerpt that captured my thoughts:

“And the single most important thing to me in the stock market, for anyone, is to know what you own. I’m amazed at how many people own stocks, they would not be able to tell you why they own it. They couldn’t say in a minute or less why they own it. Actually, if you really press them down, they’d say, “The reason I own this is the sucker is going up.” And that’s the only reason. That’s the only reason they own it. And if you can’t explain – I’m serious, if you can’t explain to a ten year-old in two minutes or less why you own a stock, you shouldn’t own it. And that’s true I think of about 80% of people that own stocks.

And this is the kind of stock people like to own. This is the kind of company people adore owning: it’s a relatively simple company, they make a very narrow, easy to understand product. They make a one-megabit SRAM CMOS bipolar RISC floating point data I/O array processor with an optimizing compiler, a 16 dual-port memory, a double-diffused metal oxide semiconductor monolithic logic chip with a plasma matrix vacuum fluorescent display. It has a 16-bit dual memory. That has a UNIX operating system, four Whetstone megaflop polysilicon emitter, a high bandwidth (that’s very important) 6 gigahertz double metallization communication protocol, an asynchronous backward compatibility, peripheral bus architecture, four-way interleaved memory, a token ring interchange backplane, and it does it in 15 nanoseconds of capability. Now, if you want a piece of crap like that, you will never make money. Never. Somebody will come along with more Whetstones or less Whetstones or bigger megaflop or a smaller megaflop. You won’t have the foggiest idea what’s happened. And people buy this junk all the time.”

This resonated with me. It’s a positive feedback loop. People are captured by their imagination, and invest accordingly. Companies sell that vision, and are incentivized to morph their sales pitch into something that is congruent with the investors’ fantasies.

“Will this AI solve world hunger?” - The investor asks.

“It really has the potential to do that!” - The company executive answers.

And this is how the narrative begins to be wove. I want to be clear, I am not suggesting that all companies are bad actors. However, the companies that actually are, have become so good at being purveyors of highly-appealing fantasies, that they have managed to skew expectations over the entire sector. Often, the result is that investors’ sky high expectations subsequently result in them being bit in the tail. As companies fail to materialize on their promises, investors are dumbfounded and in disbelief ask “How did this project fail while they were using their AI?” This is the cue for the skeptics to come onto the stage jeering, “See! I told you it was all hot air!”

This is not only harmful for investors, but also for the future of AI itself. Ultimately, this false AI fantasy has developed into widespread misconceptions about AI, both negative or positively skewed. On the one hand, there are the proponents of AI wizardry selling the magical solution to every problem. And on the other end of the spectrum are emboldened skeptics who are steadfast in their refusal to even consider the possibility that AI can provide positive benefits. It ends up being like two people who each have crashed into the ditch on opposite sides of the roads justifying the path they took because they are not in the opposite ditch like that other poor sucker. And even as each finds themselves in the ditch, we often continue to buy our side’s explanation. Why? Because that is the power of the narrative.

I suggest we pursue a sensible middle course for understanding AI that is capable of actually staying on the road. AI is neither the wizardry solution some people claim it to be, but also is far from useless. The truth is found somewhere in the middle.

One of the fields where AI has recently made its entrance, that is of particular interest to me, is in mineral exploration. Mineral exploration is a domain that is notorious for having an overabundance of data, with only a fraction of which is regularly used. On paper it seemed like the perfect environment for an “AI dowsing rod”. An algorithm that scours all the available data and deduces where the next big gold deposit can be found. Some of the first movers in this area had some major successes early on, and frankly, those early successes using AI, worked like catnip for wild-eyed investors looking for the next big thing. As they often do, they extrapolated the early success as the start of an undeviating course leading to the digital manifestation of King Midas. Every AI target was destined to yield core samples replete with gold. Unsurprisingly, the euphoric AI bender didn’t last long. As core samples of pure gold didn’t turn up in the follow up projects, many investors, myself included, were dumbfounded. But rather than blaming the company for failing to live up to my lofty and unrealistic expectations, I turned to myself.

I realized that I, like many other investors, had interpreted the idea of AI in mineral exploration all wrong. It was never supposed to be a magic dowsing rod, and neither could it function as one. There are multiple reasons as to why so, among which are the immense variability in geology, and the probabilistic nature of ore occurrence. However, as I took a closer look, it allowed me to see how AI could actually be of incremental benefit, with each incremental improvement building on the prior gains. So, the AI was a very powerful tool, just not in the way I expected it to be.

Mineral exploration is all about where to drill. As a mineral exploration company, you want the absolute best bang for your buck, and that means that a drillhole should result in as much information about a geological system as possible. Misses are costly and the margin for error is slim. An untimely or repeated miss can be the difference between a “multibagger” and vs a bankruptcy filing. It is of no surprise therefore, that executives want to be as confident as possible when delineating a drill target.

When a company is figuratively putting all of its chips on one number, additive incremental probability improvements may not have the sex appeal of the AI fantasy, but frankly, it will be an AI reality that any wise executive will jump on…

In my follow up to this article, that I will be publishing soon, I will be doing an examination of the characteristics that allow AI to stay within the lines of the road, while seeking to maximize its effectiveness and utility in the mineral exploration domain. Stay tuned.

If you enjoyed this article - please consider subscribing to my Substack: where I will be posting a monthly article in a similar fashion to this one. It would greatly help me gauge interest in how many people appreciate these articles. Subscription is free, and only consists of Substack giving a notification in your email inbox everytime a new article is published by me.

Last but not least, I want to sincerely thank @Snidely for reviewing this article and providing valuable input and corrections where necessary.