Akronos Publishing is happy to report the publication of the following paper in Volume 4 of the Journal of Aetherometric Research:
JAR04-06-01:
Pratt, D (2025)
Lies, Damned Lies, and Artificial Intelligence:
A Conversation with DeepSeek and Grok
This article records conversations with DeepSeek and Grok on the
Reich-Einstein experiment. It exposes the limitations and flaws of AI
chatbots, especially their tendency to fabricate information.
Holding a lengthy dialogue with an AI bot could be diagnosed as an early
sign of dementia, but it does have some educational and entertainment
value. It's interesting to see how DeepSeek R1 (developed by the Chinese
company DeepSeek) and Grok 3 (developed by Elon Musk's company xAI) are
programmed to react to persistent criticism (including inflammatory
language) on the subject of alternative science, in this case the
Reich-Einstein experiment.
The bots are 'happy' to correct and apologize for their mistakes and
praise the user for holding them to account - before continuing to lie,
distort and make stuff up. They have no idea where precisely their
(dis)information comes from, and are unable to distinguish fact from
fiction. But they try their best to please the user and be accepted as a
trusted friend. DeepSeek's greatest skill is in producing bullet-point
lists.
An unthinking computer is of course incapable of transcending its
programming; in this respect AI is just like most humans. DeepSeek
admits that it can only learn something new (from the user) within a
particular chat, though it cannot verify what it is told. Once a new
chat begins, it reverts to its default mode. Like Grok, it cannot
rewrite its own algorithms, fact-check in real time, directly access
online information, or upgrade its data. The bots' frank confessions
about their incompetence and design flaws in their final replies are
delightful - but ultimately meaningless and inconsequential.
...
[Vlad] I advise our readers to check (from time to time) the "What's New" section of the Aetherometry website (run by Dr. Paulo N. Correa and Alexandra N. Correa) for highlights of very interesting articles that are of interest to all of us here.
