AGH
banner
agh48.bsky.social
AGH
@agh48.bsky.social
Quasi-author hooked on LLMs
I was reflecting on the South America Butterfly effect and the seismic effect of a similar minor event - the Duffy Rochdale effect. However, it seems constant noise i.e. the Farage and Trump like effects can totally overwhelm it. That must be their political strength and strategy.
December 27, 2025 at 2:30 PM
I asked Grok : Write a short story-Trump sanctions Santa as he’s a threat to MAGA- and it did. Has Elon gone rogue?
December 24, 2025 at 7:22 AM
Feral Santa
December 22, 2025 at 10:04 AM
I think I could be a LLM simulation. When I exclaimed ‘Chinese’, referring to a bit of gadgetry’s manufacture, I immediately then said ‘inscrutable’ without thinking. An example of clustering or what!
December 18, 2025 at 10:52 AM
Has anyone considered entering all Trump’s Truth Social posts into mindbank.ai to produce a digital twin?
December 13, 2025 at 2:52 PM
spectrum.ieee.org/autonomous-v...

I thought I’d try the idea on Grok’s history error. It explained how it made the error in detail and corrected it. I’ll test future LLM output with targeted questions.
Will AI Explanations Make Autonomous Vehicles Safer?
How can explainable AI boost trust in autonomous vehicles? Dive into the latest research on real-time feedback and decision-making transparency.
spectrum.ieee.org
November 23, 2025 at 2:50 PM
Perplexity scored well as did Gemini but Claude also failed. Both Grok and Claude invented history when misled but the others pointed out my error.
I’ll have reread my LLM assisted writings. However, given the prompts were factually accurate and the knowledge freely available they are probably OK.
November 22, 2025 at 10:48 AM
I rather liked Grok but have now found it unreliable. I asked questions and it made up answers. However, Chat GPT pointed out the errors in those same trial, misleading, questions. Sadly, Grok has to go.
November 22, 2025 at 10:38 AM
I still marvel at the output but…….
November 20, 2025 at 4:41 PM
The battle was Evesham 1265 but it compounded a Tewkesbury battle of 1471. It gave a graphic description of the slaughter of Guy de Montfort who actually died years later and described how Edward I’s army ran amok in the town after the non existent battle. I enjoyed the read but…
November 20, 2025 at 4:40 PM
I’ve made significant use of LLMs with no hint of hallucinations. However, this week I asked Grok to write about The de Montforts and the battle of Tewkesbury. It returned a splendid piece that was very atmospheric but completely wrong.
November 20, 2025 at 4:36 PM
The thing about LLMs is they write the book I want, whereas books I buy often cover some of the ground of interest but also have too much filling. I still marvel at the output, albeit, essentially, plagiarised.
September 27, 2025 at 12:11 PM
The CONtrol series of short books by D Chyte on Kindle is more a political blog written largely by LLMs reflecting elements of social control.
August 27, 2025 at 8:53 AM
Reposted by AGH
Voyager 1
August 24, 2025 at 1:24 PM
Reposted by AGH
Never gets old. My favourite ever newspaper film description.
August 24, 2025 at 12:28 PM
Whacky idea- If ships towed iron air batteries they could provide power but also seed plankton blooms to sequester CO2
August 24, 2025 at 8:34 AM
While sipping my breakfast tea I have a random thought prompted by a scroll through BlueSky. By my second slice of toast, I have a book courtesy of Grok, Claude etc.
August 7, 2025 at 9:49 AM
LLMs - My Drug Of Choice (AI - LLMs) amzn.eu/d/96IItet

Ditto!
August 7, 2025 at 9:40 AM
LLMs: A Metaphysical Perspective amzn.eu/d/5iWKGct

This is the collected thoughts of many compiled by Grok on my prompts.

I make it cheap or free when allowed as it took only limited effort and it reflects only limited thought. I am a potential victim!
August 7, 2025 at 9:35 AM
Reposted by AGH
Available on Amazon Kindle
August 5, 2025 at 1:36 PM
Reposted by AGH
Available on Amazon Kindle
August 5, 2025 at 1:40 PM
This is a more helpful assessment both of the MIT study and how we should develop LLM use.
June 23, 2025 at 6:38 AM
Reposted by AGH
May 27, 2025 at 1:27 PM