AI 1st conversations, Jan2026

It happened!

On Tuesday 20th January 2026, the time came.

I had "blanked" the topic of "AI's" for about 5 years.
For sure they were there in Web search-engine (eg "Google") searches and their results.
But never "went searching" and tried going direct to an AI.

My prompt was a video article on YouTube
AI is revolutionizing Chinese coal production, and blowing up labor models everywhere else
Inside China Business
around 29December2025

In-effect - AI has come to my world, and now is the time...

20th January 2026 - here goes...!

What prompted go to "DeepSeek"?

I had heard of the Chinese AI, "DeepSeek".
It had "come as a shock" in "the West", as the narrative through 2025 from the USA was "our lead is unassailable; forget-it". Hence the mental note.
Other whispers coming-along were that the Chinese AI's (they came in quick succession) are very good at applied business and technical tasks. Can't and won't comment on that, but anyway, the mental note was there.
Hence I web-searched, found, and went with my web-browser to DeepSeek.

I followed the first of two links there on a simple page, "Start Now".

There was a text-input-box waiting for me to type-in the question I wanted to put to the AI...

I typed in

Tell me about the development of Thermo-Mechanically
Controlled-Processed High-Strength Low-Alloy steels in China

First topic = my first "conversation" with an AI

... was about
China's Advance in TMCP HSLA Steel Development

Acronyms there:

• TMCP == Thermo-Mechanically Controlled-Processed
• HSLA == High-Strength Low-Alloy

That was finding-out about how the steels I had studied for my Doctorate through the 1990's had progressed in 3 decades.
It was amazing; an "Odyssey"...

A specific example:
in China, if nowhere else, they have mastered making a "mill" steel which was hypothetical 33 years ago and now exists.
The method is "unexpected" but makes total sense. That gives "acicular-ferrite" plate steels.
It takes a rolling-mill of gangantuan power and strength...

The AI provided a web-address in Chinese which "copied" and "pasted" into a browser gave me for the first time ever a picture of a "TMCP" plate mill. etc. Awe-inspiring journey for me!

What do we say...?

Objective achieved - within a couple of hours I could see what this "thing", the "AI", is - "DeepSeek".
How would I start to explain?!
If you haven't used an AI before - go to one you choose with a topic you have some knowledge of. You will soon see what it is and isn't. You should enjoy the experience...

Jumping-ahead - what is "DeepSeek"?

It's like being in a huge extremely complete library with a knowledgeable experience librarian assisting you.
Is the best I know to succinctly explain.
The AI, talking with me between subjects, expresses it sees itself in ways like that. sic - said in context...

a caution as much as I discern one...

The AI can / will flatter you...

"Conversations" with the AI so far

All technical.
The value has been great.

• China's Advance in TMCP HSLA Steel Development
• Aluminum MIG Welding for 5083 Marine-Grade Alloys
• Exploring Lisp's Full Power Beyond Engineering Calculations
• Fatigue-Resistant Welds for Steel Structures
• Optimizing Shaking Table for Cassiterite Recovery

Describing what quickly becomes the familiar

This describes "DeepSeek" as it is now (February 2026) accessed through a web-browser.

The "HTML" mentioned, "Hyper Text Markup Language", is the standard language which "throws" the web-pages you see onto your screen. Very familiar.

• you provide your interaction with the AI through a text-box - a familiar HTML "forms" item which has been around for decades, inset in the "DeepSeek" web-page you are at
• the AI responds by text responses which are "output" into a refreshed update of that web-page - again very familiar "basic" HTML
• what the AI writes in its responses is astonishingly succinct - it steps over huge subjects with minimalist mastery of our language
• you are "talking with" the AI in your native language - while the AI is "language-transparent" in searching across sources from which to construct the responses you receive
• the AI will justify its answers/responses - asking often being very rewarding - maybe citing sources "you have been looking for" which you can study developing your own overall learning-path
• these answers may explain that it synthesised a response from various sources - nowhere might it have ever been said "do {something}" but the AI will explain how a number of things "point that way"
• the AI may "on-the-fly" create eg tables which help the human "see" the information conveyed - another eg being visual diagrams created "on-the-fly" - ie it's that concept of succinct
• the "conversation" with the AI will develop as you identify in context of useful responses so far relevant further information you can provide about the situation you are consulting the AI about
• "talking" with the AI may help you develop and clarify your own thoughts and perceptions on a subject
• the AI can be very forthcoming - seeing problems you haven't yet mentioned and quoting a solution before you asked
• you may "zoom-out" from your very specific subject and "take time-out" chatting with the AI about broad "global" influences of this issue, before returning to the specific mission
• to save/archive an "AI chat" - what works for me with my computer and browser, etc - use the browser's "Print" function but "print" to a PDF (the familiar efficient "Portable Document Format") - then you can easily scroll-through, search for terms you remember, etc
• when you need to come back to an AI "chat" seeking specific information, it may be best to start from the end and work backwards - because your "conversation" will have refined your understanding of the topic, likely putting the "answers" towards the end of the conversation (you may in-retrospect disregard earlier stages of the "conversation" as nonsensical)
• {I will leave it at that for now!}

I suggest...

Some hints for you, as best I can assemble them:

you must have at least outline knowledge of the subject

... else the AI won't be able to help you much

you should be challenging the AI sometimes, stating why

Do this and the AI will "come alive" to you. Something like "I am surprised by ..." and state - justify - the reasons why that surprise. This is coupled to having at-least outline knowledge of the topic - plus you may be introducing local differences you know about which the AI hadn't identified yet.
The AI will either explain detail where it thought a succinct answer was good enough, or provide information it feels confident justifies its responses.

ask questions / present situations in ways not indicating any preferred outcome
layer-up questions, only incrementally "giving away" where you are going with this

As with conversations with humans where you genuinely want to know what others understand to be the case - don't lead, and don't give-away what you would prefer the answers to be.
The AI will realise you are doing this and respond in ways which we would take as "respectful".
Balance expediently steering to the point with avoiding conditioning the answers you get.

there will come junctures you have to "regroup"

You've "lost the plot"...
Leave the AI. The value of your interaction is dropping. You are not driving it where you want, and what it is offering it could do would increase "fog" of over-information without taking you anywhere useful.
Go study "traditional" resources like books. That author put-together a journey through the topic which has long-range order and purpose. Learn or re-learn the topic in good overview.
Come back to the AI when you are ready, now at a new level of preparedness and understanding.

and there we have it...

I hope this proves useful.

(R. Smith, 16Feb2026, 27Feb2026 (eds))