Examining Tony’s Open Chain: Is This a Good Use of AI? | #PodSaveChocolate Ep 126
Episode 126 of #PodSaveChocolate features an exploration of using Gemini to summarize Tony’s Open Chain Report for the 2023-2024 reporting year. Is this a good use of AI? What can we learn from Gemini’s analysis? What are the limits? Is anything missing? Can Gemini be trusted? [updated]
When and Where to Watch
Links below to watch LIVE and to view the archived episode.
Click on this (shareable) link to watch on YouTube. Please subscribe (free!) to the @PodSaveChocolate YouTube channel, like this video, comment, and share this episode to help grow the #PSC community.
Watch and comment LIVE or view the archived episode on LinkedIn. Join my network on LinkedIn to receive notifications and to refer business to each other.
Watch and comment LIVE or view the archived episode on TheChocolateLife page on Facebook (for 30 days, then watch the archive on YouTube).
Follow TheChocolateLife on Facebook to receive notifications and catch up on other content.
Episode 126 Overview
A prior episode of PodSaveChocolate focused on Tony’s Open Chain. In that episode, I mentioned I had used Google’s Gemini AI tool to do some “Deep Research” on Tony’s Open Chain Report for the 2023-2024 reporting year.
Clay Gordon
You can download the all of the Open Chain reports from this page.
But, before I jump into the main topic ...
On the Saturday before the Tuesday this is going live, it occurred to me to wonder if I could trust one of the graphics in Tony’s 2023-2024 Annual Fair Report (AFR).
Specifically, the graphic on P167.
It turns out the numbers don’t add up and the stack plots aren’t drawn to scale.
The text in the KPI says “The amount of premium paid [per tonne ] to reach “the [LIRP 1 ”]. Also, note that while the €uro amount of the premiums is listed, there are neither totals nor percentages shown. (I added those in – with caveats.)
What the LIRP was during any of the four crop seasons shown is not mentioned on this page. ↩
Note that the LIRP is not represented along the vertical axis of the graphic nor is there any separation between the 2022/23 numbers and the 2023/24 numbers 2 .
I mention this because the 2023/24 LIRP may have been different from what it was in 2023/23, and neither are shown on the page for context; a critical datum is missing, making the 2022/23 numbers not directly comparable with the 2023/24 numbers, which I am confident is part of the design brief.
“If you can’t dazzle them with brilliance, baffle them with BS.” 3
So I added horizontal reference lines and numbers to the vertical axis and added vertical lines to separate the 2022/23 numbers from the 2023/24 numbers. ↩
Widely attributed (but now considered to be misattributed) to William Claude Dukenfeld, more widely known as W.C. Fields. The full quote is anecdotal and is attributed to a “Bill” in a larger conversation, so William/Bill may be the source of the confusion. ↩
My annotated version of the page. This is v2 – the vertical line separating the 22/23 from the 23/24 years was originally in the wrong place.
I used Photoshop to do the annotation. I calculated the height of the €2200 LIRP line after counting the number of pixels from the €0 baseline to the top of the CdI €2268 farmgate price bar to establish the scale.
The CdI main-crop 2023/24 numbers total €2200, but the plot falls below the €2200 reference line. The CdI mid-crop 2022/23 numbers total €2198, but the total height of the stack is greater than the height of the main-crop 2023/24 stack.
From this one example, it’s clear to me that NONE of the graphics in a similar format in this AFR should be trusted to accurately depict the numbers that are used to generate them. This also warrants taking a retrospective look at every graph in every AFR.
Now back to the main point for the episode ...
I uploaded just the 2023-2024 Open Chain report and asked Gemini one question:
It’s nice to know that Gemini’s input parser is tolerant of misspellings – deliberate in this case.
The rest of this episode is going to explore the results of Gemini’s “thinking” by evaluating the report, which you can download (MS .docx).
Additional Links and Resources
Clay Gordon
The CSS for the embedded file was interfering with the default template CSS so I moved the report HTML to its own page.
TheChocolateWire
2022: Uncommon, Taza, Dandelion, Tony’s, Chocolate Scorecard
TheChocolateWire
Questions?
If you have questions or want to comment, you can do so during the episode or, if you are a ChocolateLife member, you can add them in the Comments below at any time.
Episode Hashtags and Socials
#AI #Gemini
#Tonys #TonysChocolonely #TonysOpenChain
#cocoa #cacao #cacau
#chocolate #chocolat #craftchocolate
#PodSaveChoc #PSC
#LaVidaCocoa #TheChocolateLife
Future Episodes
#PodSaveChocolate and #TheChocolateLifeLIVE Archives
To read an archived post and find the links to watch archived episodes, click on one of the bookmark cards, below.
Clay Gordon
Clay Gordon
Audio-only podcasts
Become a Premium ChocolateLife Member!
| These offers are available to free members, so subscribe above then click one of the following links. | |
|---|---|
| Team TCL Member Monthly membership | First 2 months FREE (save $10/yr) |
| Team TCL Leader Monthly membership | First 2 months FREE (save $30/yr) |