Watch BingGPT go Germanium Hog Wild!

Here is a short wire for those on a small research budget who like to have fun.

When the whole xGPT thing first broke I was a bit perplexed about how to apply the stuff in my Jevons Global research workflow.  That did not stop me having fun.

You may recall the infamous PopTart Conspiracy Theory generated by ChatGPT.

To test out the MegaBrain AI, I sent this query to ChatGPT:

Develop a Pop-Tart Conspiracy Theory on why Pop-Tarts always wind up burned on Sunday morning when the toaster is left in the hands of Ageing Boomer Grandpa who can't work his smartphone.

This generated the result:

The Pop-Tart Conspiracy Theory suggests that the burn rate of Pop-Tarts on Sunday mornings is a deliberate effort by the Pop-Tart Corporation to drive sales. The theory posits that the company has embedded a subliminal message in the packaging, instructing Grandpas to set the toaster to the highest setting, resulting in burnt Pop-Tarts. This, in turn, prompts frustration and a desire to purchase a new, high-tech toaster that won't burn the beloved breakfast treat. The conspiracy theory suggests that this is all part of a larger marketing strategy to drive sales of both Pop-Tarts and high-tech toasters.

Yeah. Idiots like me could waste hours on a trading desk doing that.

However, I also have to run research, and an investing business, on a very low budget.

More than ten years ago, I set up my business. Having left a big investment bank, I believed that I could use new technology to achieve (at least) a 10x scale-up in research productivity.

Initially, I went pretty hard on doing everything myself. Ran an on-premises supercomputer, fiddled around with GPUs, ran web crawlers, the whole box and dice.

Doing that is a great way to learn stuff, but not a great way to make money.

Venture Capital does not back older folks who like to play with hardware :-)

However, now that the AI is in the cloud, and very readily available, it is fun to see how you might crank it up and hit the 100x to 1000x productivity target, ahead of my initial 10x stretch goal.

That now looks quite doable, to me.

On the trail of Gallium and Germanium...

I wrote a little while back on the Chinese Gallium and Germanium export controls. The whole thrust of that note was to point out the critical role played by smelters. 

In the contemporary financial market, there is almost never any detailed research so I can get away with stuff. I can say things like...

Of course, I laugh when I read of Australian miners going off to explore for Germanium. Good luck with that plan. 

and get away with it.

Mostly, that is the right attitude to take to this small global market.

However, it also helps to study some history, namely that of the Tsumeb polymetallic deposit that served the global germanium market back in the 1960s. 

Australian polymetallic explorer Golden Deeps (ASX: GED) recently announced some promising hits for both Germanium (Ge) and Gallium (Ga) when they re-examined one drill core (NSBDD008) at their Nosib Prospect in the Otavi Copper belt of Namibia.

This is not that far from the famous Tsumeb Cu-Pb-Zn-Ag-Ge-Cd mine that closed in 1996. This used to be a significant source of Germanium, until secondary production from Chinese zinc smelters pushed prices too low for many operations. If you smelt lead and zinc in truly massive volumes, then it can be difficult to compete, even with higher source grades.

That is the reason for my flippant remark on primary germanium or gallium mines.

It could happen, but you would need something special like Tsumeb to make that work.

Some fun with BingAI Search...

Now that I actually know of the Golden Deeps Nosib Prospect, and where it is located, I am led to think that some better search tools could have helped me spot this sooner.

Out of interest, I ran a little experiment with Bing AI Search to see how well it did.

These are current experiments for me since I do a lot of work for clients in helping them to refine their thematic investment portfolio construction. I am doing one now (not on this).

Here is the first experiment I did in an open Bing Chat window.

When I gave Bing Chat the Golden Deeps URL, it happily goes off to give me a summary of the business activities, and with a little coaxing, gives me a JSON package back.

It nails the most important commodities, namely Vanadium (V) and Copper (Cu).

Since these chat systems remember context, I can now vary my questions a little to zero in better on what other things I might learn from the website.

This is pretty good, for human conversations about just a few websites at a time. However, Jevons Global is a two-person company. We need code-driven queries on thousands of different companies, not just a short-list of a dozen to query in one afternoon.

You can see that the query above is making progress in populating a nested data structure that would be more useful for scraping many websites.

However, the other thing to consider is how to get very specific information back when I know it is there, but I cannot be bothered to go back to the website.

This is the whole reason to mention the Germanium hit for Golden Deeps. I already knew that they had one but could not remember in what context I read it. So, I made this query:

Does Golden Deeps have any Gallium or Germanium at any of their projects? If so, name the project, where it is, and give me the ASX announcement date and URL.

Sneaky, huh? Lazy too. I like sneaky-lazy the best.

How about that?

In case you are wondering, the whole reason I did this experiment was to see if Bing Search was good enough to give me back the ASX Announcement URL in a useful form.

{

"Nosib vanadium-copper-lead-silver project": ["Germanium", "Gallium"],

"Announcement date": "07 July 2023",

"URL": "https://www.goldendeeps.com/wp-content/uploads/2023/07/HighValue_Germanium_and_Gallium_Identified_at_Nosib.pdf"

}

That weird looking stuff is machine readable JSON, which is what my computers want.

Noice, as they say in Fountain Lakes...

A girl could get very effluent if only she found enough Geranium down the back of the sofa.

Move over Barbie...

Investment conclusion

Of course, to make this work well you need to automate that workflow in code.

I did that already, by going here. That approach was free (not in time - just dollars).

However, you should not do that as I am a shareholder in Microsoft (NASDAQ: MSFT).

You should go here and pay for it!

Since Jevons Global has Microsoft as our biggest position, you should own some too.

Own the stock and pay for the AI Search. That is my considered recommendation.

It may also be worth looking into Golden Deeps (ASX: GED).

This question says it all:

What metals has Golden Deeps found, and name which project they were found at?

The answer is interesting:

Golden Deeps Limited is an Australian mining company that has identified the chance for early cash-flow from its 100%-owned Abenab vanadium, lead and zinc project in NE Namibia after testing showed that existing above-ground stockpiles could be processed using simple gravity separation. A final bulk concentrate sample yielded 8.9% V2O5, 30.5% Pb and 8.95% Zn.

The Bing Chat system does provide references.

With a bit of extra work, we find that the concentrate data is sourced from here.

What is important to note with these systems is that the above news story is actually pretty old. The date and time for that information is 22-Aug-2019 at 17:27. It could be stale data.

Of course, prior to making any investment decision, it is vital to go back to the primary sources and work out the current state of operations, and the actual progress made on the project.

Never miss an insight

Enjoy this wire? Hit the ‘like’ button to let us know. Stay up to date with my content by hitting the ‘follow’ button below and you’ll be notified every time I post a wire. Not already a Livewire member? Sign up today to get free access to investment ideas and strategies from Australia’s leading investors.