Andrew's AI crash-course

New try: look also for AI-assisted story-telling tagged below

One more for the road: Andrew Zolnai Private Blog posts

Last update: by the numbers at bottom, with a dash of humor

And another one: last Kuwaiti Blog post with new experiment here

Closed: 'AIding' Tool, above videos & at bottom

Updated last Medium Pro. post just above Medium Personal

Added: AI generated GIS maps, under Flickr

Added fact-checking tip & shortcut under: AI's little helpers, halfway down

New channels: Kuwaiti blog, YouTube Shorts and Other

New how-tos: building AI agents + disaster response

New human-centeredness: with a little help from AI

New tools: Copilot help & Perplexity summaries

Update 4: a three-part manifesto closes this review

Update 3: added a note on "AI's little helpers"

Update 2: a French look into the future of AI

Update 1: ongoing collection of AI mods on Flickr 

Have I mentioned that AI will need skilled writers more than skilled coders? Forbes wrote here about a US efforts in that regard. [ 6 mo. update: Fortune reiterated that here. ] The one thing AI won't replace,  acc. to the starred "AI-ding Tool" example below is curiosity:  it mirrors Steve Jobs' "stay hungry, stay foolish" at Stanford 2005 Commencement Address here.

Exhuming my thesis, part V... with two sides of AI!

 Update 2: see a second side of AI here (although I did quit AI & socials, see my last post)

Update 1: another GIS also has an AI offering I tested below

This follows on Part IV with further AI investigation in GIS that created the maps in previous Parts. This is also an update on my Community Engagement series ending here. Our non-profit has an opportunity to re-engage GIS & AI after my partner attended an AGI meeting on same. I plan to repatriate to pursue this, stay tuned.

GIS:

We decided to pursue AI offerings on QGIS as it's free - my other account off maintenance curtails my access to AI tools - and the prognosis looks good. First I did a review of what's available in Google Gemini here. The update esp. the closing caveat were particularly useful.

click to enlarge

Here are the latest AI plugins on the latest version of QGIS.

click to enlarge

I loaded my last project I used in QGIS to review my thesis 40 years on, starting here & updated right here. The integrated chatbot question is at lower left, and the answer at middle right confirms that I have used the appropriate data set for regional geophysical data to study geologic terranes in that area.

As with AI topics in my blog feature post here - see this here posted ¼ way down - let me update you on GIS & AI as I keep exploring... Drop in comments below where you think I should use AI in GIS?

Update

giscloud.com proved useful to bird-dog my decades-old projects; I started using it 15 yrs. ago near their launch, because they were the first on-line vector-based GIS... now they prove their ingenuity again w AI baked-in, not a side-car as in QGIS!

click to enlarge

New world & old world

I happened to reread "Whence the Mountains" (Google Books), a memoir in honor of my thesis sup who launched me on this whole adventure: reference is made into a college essay I asked AI to find (Perplexity); not only did it find it, AI suggested how & where to source it... Lo and behold! University of Manitoba Archives sent me a copy barely over hour later... now that's what we call service!

Isn't it also a cool crossing of old & new, AI sourcing  from a college magazine the "words of the master from 72 years ago", per memoir editor?

click to enlarge, full, archives@umanitoba.ca

Read to the end below to get the crux of the issue. And if you read the last two poetic references, you get a hint of one of Ray Price's last piece on the geo-poetry inspired by his mentors SW Carey & WH Hamilton. His 2012 Penrose Gold Medal talk he concludes with a  "geo-poetic conceptual model". Warning: it's long and technical here; thanks co-alumna Tekla Harms for nominating him here.

click to enlarge, from above

Personal footnote: I experienced a similar adjustment in geologic context - although in using & not creating ideas - in my last undergraduate year at the University of Calgary, I attended a plate tectonics class with JF Dewey and K Burke. They inspired here my 4th yr. paper on aulacogens (Wikipedia). In my thesis area I realized the Southern Province was missing the symmetrical counterpart to the S - aulacogens like geosynclines were old concepts assuming symmetry, the absence of which led to drop the former and to see continental margins as extant halves of the latter - this led me to surmise a Manitoulin Island Discontinuity ending the original offshore extent in my thesis, review starts here. 

AI Prompt Engineering Trial 2

Community Engagement 1,... 20, 21, 22 & 23

Update 2: next blog installment is here.

Update1: Medium prof channel reflects this here.

My post here, starting my experiment to convert my CIC (non profit) from Cottenham.info to Cottenham.ai, showed how can we use AI (Wikipedia) for good. My CIC started to address rural isolation in East Anglia in light of climate change and the pandemic (here and here). After a COVID hiatus forced this reassessment, the advent of AI piqued our interest beyond the hype in the press - see my companion Medium series on this topic, following four posts from The future is digital… only it’s not what you think! here - this lead into a curious interaction and my partner finding a new direction.

"AI for the rest of us", Part II

Update: follow on Part III.

 My previous post (here) showed a useful if trivial example of Bing AI (here), not only replying in the addressed language but also quickly & easily teasing a relationship among related items.  Now let's look at a non-trivial common task, to summarise a 110 page document.

Last post: views since Sept. 2009 top a million

Update: the last three posts below picked up further AI using GIS, AI to convey complex astronomy and global-to-local integration efforts re: Climate Emergency. See my last two AI posts on companion blog here closing my social commentary launched in Kuwait. My Medium Professional Channel also closed on a humorous take on AI here, & my Personal Channel did here. Note also "last post" entries in my feature post atop my blog here. Note finally my internet footprint on the banner menu of this blog's web view here... 

Update 2: at bottom, make that 1¼M... six months later! 

Update 3: And exceeded 10K hits per mo. a month later...

March 2026 milestone

Is reaching this milestone not fitting on the blog's 15½th anniversary? The 20th anniversary of my return from the US, a mo. to the 50th of emigrating to Canada, & 6 mo. to the 40th of my original pre-internet post ( see 1986, top left banner menu of this blog's web view here)?

As cumulative hits exceed the 1M mark (see 6 mo. update at bottom) let's update the last recap here. 

click to enlarge (see note at bottom)

This is how I ended my feature blogpost on AI:

September marked 15 years of blogging: quite happy with 375 posts garnering over 750K hits topping 50K monthly, up from under 10K before addressing AI; see my next post for stats in the context of web history incl. mine. My web footprint is on the banner menu of this blog's web view, where you can see hits at top right. 

Update: March 15 2026 exceeded 1M hits at 400 posts with monthlies nearing 50K hits. Note that Sept. 2020 had ½M hits... so hits doubled in last 5 out of 15 yrs.!

May 2026 detailed evolution per Copilot

I tabulated monthly hits since the beginning here, and the pivot table (Wikipedia) summarizes 15½ yrs in tabular and graphic form:

click to enlarge

click to enlarge

Copilot AI traversed the 15½ yrs to tabulate its themes as follows:

Copilot synthesizes it into a Emergence Timeline:

Click to enlarge

Copilot estimates a qualitative Intensity Timeline:

click to enlarge

It creates this Visual Temporal Graph:

click to enlarge

Footnote: is it not amazing how Copilot suggested the method, ran the Python script based on the blog to post this graph... all without writing a line of code?!

And finally we have a Color‑Coded Theme Matrix (ASCII Heatmap):

click to enlarge

🧭 What this matrix reveals:

1. GIS dominates early, then stabilizes

    A strong red block on the left, tapering to yellow — your technical foundation.

2. Climate peaks sharply in the middle

    A bright red column in the Climate Era — your environmental deep dive.

3. History explodes suddenly

    A dramatic jump from green to full red in the Medieval Era.

4. Personal + Narrative become your late‑era engines

    The Integrative Era glows red at the bottom — your reflective, essayistic turn.

5. Tools fade but never vanish

    A warm orange → yellow → green gradient — always present, never dominant.

________________________

Apr. 9, 2026 update:

The hardest problem AI ever solved, with Google Deep Mind CEO

In the 1st ¼ hr., I get vindicated for putting my work on the web free-to-reuse (called CC BY-SA 4.0 at the bottom of this blog's web view): AlphaFold was put on the web for free & pretty well every drug is now made today using it (~ 13:00 mark). See also their full movie at the bottom of my feat. post atop my blog. Aside from the million hit mark above, this is a nice way to wrap up my "the internet is free" socials, isn't it?

Update: ditto, bottom of "AI-ding Tools" here, "How Nature Solved Robotics" here ends up sharing widely the know-how around DIY robotics to help improve collective learning... I rest my case!

Last one: posting math problems on the open web spurred pre-AI as well as AI mathematical endeavors inside academia and out: Why the Legendary Erdős Problems Are Falling to AI - Quanta Magazine 

Sept. 15, 2026 update:

My blog hits show no sign of slowing down... 

click to enlarge

Note: the last "drop" is an artifact of counting partial months - compare to the top graph & see numbers jump right away.

Note also monthlies (seen on web not mobile view here) rocketing from ~5K toward 15K... 

Closing addition: see at the bottom of the feat. post on AI here the cumulative graph that echoes the sentence.

"AI for the rest of us", Part IV

Update: follow on Part V.

 The previous post (here) recapped our purpose to use current lessons-learned in new tech to help our community engagement. Here is another way to use AI, to summarise and to decant - summarise in a structured manner - information from an article my colleague Terry Jackson at Cambridgeshire.ai (under construction) asked me to try using Google Gemini (formerly known as Bard, their AI tool).

"AI for the rest of us", Part V

Update 1: go to the bottom for a positive aspect of what AI can automatically generate, and how a test for AGI (artificial general intelligence) has been designed

Update 2: see next part here of this series for an ultra-simple use of AI

While I described opportunities AI gave in prompt engineering and text processing (ending at previous post here), I tried for fun to create a billboard from an advert I saw in TX or CA 20 or 30 yrs ago: it was Home Depot's "got wood?", after the wildly famous "got milk?" campaign from the Milk Board. 

Showing folks  sporting white 'mustaches' from drinking milk too eagerly, it was the dawn of the photoshop era: milk was traced into mustaches or smiles... "and no milk was spilled in the production of the advert".

AI to recreate a lost picture

This follows on AI "lessons-learned, lessons shared", last post here and master post here.

Here is another use of Google Gemini AI. A LinkedIn discussion here described fog/cloud banks in Cen. & Nor. Cal. against nearby ridges.  I commented on same viz. Yucaipa Valley & Ridge in So. Cal. Here is my comment:

AI Prompt Engineering Trial 1

Community Engagement 1,... 20, 21 & 22

Update 2: See the next step here in experimenting with AI for good via this CIC

Update 1: See companion Medium article Stop AI scraping your internet data.

The previous post (blog) introduced how our CIC will attempt to find government source data of interest to East Anglia Parishes, in order to assist in creating web pages for all 254 of them. AI is used here to assist the task of seeking information then creating website templates over large numbers of sources to a large number of pages. This is in order to bridge the digital divide of rural citizens and their administrations, which are not plugged into centralised corporate or government ecosystems. We acknowledge the assistance of local Wikimedians (here), who are transforming voluminous IPCC (Wikipedia) reports into text the general public can grasp, retain and take action on.

How to build AI Agents

This follows on my AI "lessons-learned,  lessons-shared" mainly on my Ongoing Crash Course pinned atop this blog, that links all items on my various socials.

Maryam Maradi (LinkedIn) shared on-going how-to's I pay-forward here to the rest of the world. I posted the texts on the links below her excellent posters also in-text (click to enlarge). My (liked) comment on her second post was:

This is so excellent! I'm a geologist & GIS-er of 40-odd yrs., w interest in geophysics in general & seismic in particular. I have never seen such a comprehensive review and sign of hope for disaster response... Thanks!

"All you have to do is ask AI", sometimes

Let's follow on this post here about more intriguing geomorphology (land forms underpinned by geology), with a little help from AI in the first instance.

Hungary 

LinkedIn @markku-ylisirniö posted here a cool map of Hungary:

screen grab from original post, click to enlarge 

"AI for the rest of us"

 Update: follow on Part II.

In my other blog (here) Language and mores, Part VII (here) described what I found out about my family's status as immigrants in our complex story of a pregnant mother's escape from Hungary and my siblings ½ generation later. When I later discussed how I thought my marriage ended, my family saw a Jungian influence in my thinking. Jung's disciple Adler apparently inspired Kishimi & Koga's recent best seller "The Courage to be disliked" (Google books). Ensued a three-language dicussion to understand what this was all about... Mum prefers our mother tongue Hungarian, my brother lives in Montreal and my sister in Paris.

"AI for the rest of us", Part VII

Update: the last of a series, follow my recap blog post here incl. Medium Channel posts on AI.

Note: the last post on my personal Medium channel shows how I used Copilot AI to good effect for fact-checking. 

Last we showed here how to create "quick & dirty" graphs while chatting on ur mobile. Now we'll see how we can create summaries, not from "text to hand", but from that cribbed off the internet. 

Having left socials, I'm down to corresponding at Extinction Rebellion "Media Tell the Truth" WhatsApp group. Jon Fuller there is also at "Climate Genocide Act Now" here. I was a scientist for Extinction Rebellion, and blogged here about the pandemic against the lies of the British government, as well as about floods and rising sea levels in East Anglia. I didn't know GCAN, however, as I'm no longer active.

What is your 'ikigai'?

Note: 2nd video mentions AI well before its time...

Update: see a Perplexity AI detailed definition at bottom.

 Ikigai (生き甲斐, "a reason for being") is a Japanese concept referring to having a direction or purpose in life, providing a sense of fulfillment and towards which they the person may take actions, giving them satisfaction and a sense of meaning. en.wikipedia.org

"AI for the rest of us", Part VI

Update: Part VII here follows w using ChatGPT to summarize off the web direct.

 Part V (here) showed elaborate (mis)use of AI. Here is a super simple example whipped up in Google Gemini on the phone during a chat about value change over time. Notice all along, adding less & less detail shows AI perists previous prompts.

Community engagement and social prescribing

Community Engagement 1,... 20, 21, 22, 23, 24 & 25

This follows an update on this blog here, where I announced my move away from Cambridge. This is also a good place to end the Community Engagement series. I will however continue on topics as they arise like re: AI here, or Sea Level Rise here & updated Thames Barrier starting here. Follow also the Blog Archive to the right-hand side of the desktop version here, as well as the right-hand side of the Banner Menu atop.

We recently rebranded Cottenham.info to Cambridgeshire.ai: the domain name is acquired but page not done yet finished; we have a prospectus as an evergreen document, meaning ever evolving.

Our community involvement over 5 years in March is one object of this blog starting  here (follow the links) and listed here. We engaged with various community interest  parties at the impact of climate change then the pandemic on social isolation in East Anglia.

Cottenham Open CIC rebooted

Community Engagement  1, ... 12, 13, 14, 15, 16, 17, 18, 19, 20 & 21

Update: the next post (here) will show how a blog post on creating maps, is created by generating prompts  using Bing AI in a Udemy course on prompt engineering following STAR (style-task-audience-role).

Re-engaging après-COVID what is in LinkedIn, below & Esri Hub: Part 1 outlined Community Engagement. Part 2 built a story introducing the community. Part 3 tied together community maps and climate mapping. Part 4 introduced a process framework for this community engagement. Part 5 expanded on our aim toward a community engagement.  Part 6 added our own Wikipedia Gazetteer as we build up the local landscape. Part 7 showed a draft Press Release introducing our social enterprise. Part 8 on coastal inundation scenarios adds some parameters in the debate. Part 9 on temperature anomaly scenarios further constrains the debate. Part 10 followed up village engagement process via recent Parish Council update. Part 11 added flood risks to coastal inundation and temperature regime models. Part 12 described Cambridgeshire Parishes affected by sea level rise. And finally here we introduce AI with a local twist.

Exhuming my thesis area over 40 years later

Update 3: this post here furthers the investigation adding AI to GIS mapping. 

Update 2: this post here closes this series & loops the loop with AI to the rescue.

Update 1: see here papers a decade later that point also to pre-Cambrian terranes.

My 1982 M.Sc. thesis, Regional cross section of the Southern Province adjacent to Lake Huron, Ontario: implications for the tectonic significance of the Murray Fault Zone (ResearchGate) proposed a plate-tectonic (Wikipedia) setting around 1.9 billion years ago, in the Proterozoic (late Pre-Cambrian). I proposed there to be another buried land mass south of what I called the Manitoulin Island Discontinuity. While plate tectonics was well established in the later Phanerozoic (Wikipedia) record, it took a lot longer to be accepted earlier in the geologic time scale with lot sparser rock record (see "More" at the bottom).

"AI for the rest of us", Part III

Update: follow on Part IV.

 Part II (here) showed a small but significant use of AI in preparing our Prospectus (link in that post). Let's look at how we're taking this further now.

The Spectator did a great state-of-the-AI here, including basically what we did above on steroids. They also highlighted Google Gemini, Bard's successor we signed up for. We also joined Wikimedia UK in open data space.

New Years Eve 2000 racing by Greenwich Meridian

Update: note a dash of AI mixed in at the bottom for fun

Our TV rung in 2025 live in Paris in my time zone, and six hours later in NYC dropping its ball in Times Square; nothing matches however this cracking event in London to usher in the New Millennium: barges were aligned in the middle of the River Thames, spaced exactly 1 s. apart as the earth rotated past midnight GMT in London, from the Meridian at the Millennium Dome to the east, to Vauxhall Bridge to the west.