EPM (AI: valuation math, comprehensions, delusions and errors)

With the EU CRMA Strategic Project Status - OCT having pressure now to conclude, and waiting on one project.
I thought it might be time to set up an isolated part of the forum to tinker with AI valuations and delusional takes.
I've tested Deep Seek, Co-pilot, Chat-GPT, and thus far - Grok actually tries to work something out.
The numbers are so what excessive so I don't thinking posting $XX billion or $XXX billion maths for in-situ valuations are logical.

But it's an interesting exercise of spot the difference - same algorithm, but 'delusions' from the monte carlo decision tree, can shift matters more bullish or bearish. Note also weird uses of an explained discounting multiples sometimes used twice to damper prospects and align with peers.

AI does lie when it is forced to spit out a response - but the question is can you spot the true parts?
It's important to know what you own.

Have a read - in your own time; and watch out for the delusions, to find some truth.

Latest to oldest (done in the space of the week).

(Using a "Role" as an analyst"- the valuation process is more delusionally bullish)

https://x.com/i/grok/share/o1GWx94VV87elV8is6lav0njO

(Using a "Role" as an analyst"- the valuation process is more bullish)

https://x.com/i/grok?conversation=1927238521133121597

(No Role - task provided, Splits Tanbreez deposit up)

https://x.com/i/grok?conversation=1927236087308567016

(No Role - task provided, uses very low multiple to in-situ valuation)

https://x.com/i/grok?conversation=1927231967709237259

(No Role - task provided, uses a NPV assumption and multiple)

https://x.com/i/grok?conversation=1927230063616868614