[Stoves] What do the AIs ChatGPT3.5 and Perplexity know about the upward and downward flow of air?

Wed Feb 28 16:52:27 CST 2024

Hi Crispin et al;

The free AIs clearly still have limitations, but we can see where this is
going.  For an AI that can access the scientific literature, asking them
questions could greatly speed-up finding the key references to support a
theory.  Presently, a human has to scan through a number of papers, and
check their references to find good sources.  Depending on the topic, that
can take weeks.  The more exoteric the question is, the harder the job can
be.

I had a very useful dialog with ChatGPT the other day about the
economic impacts of social housing in Vienna, Austria.  I was able to ask a
number of off-the-wall questions, and got some really interesting answers
with things I had not thought of.  Unfortunately,  when I asked ChatGPT for
references to back-up its answers, it produced nothing but hallucinations;
papers with authors, titles and DOIs that did not exist.

One thing that my little test with up/down air flow showed is that
just like discussions between humans, the participants (asking conceptual,
non-analytic questions) will be coming from different viewpoints which may
actually be contradictory.  Just as in human conversations, we are NEVER
fully aware how our viewpoints converge and diverge.  Much could be 'lost
in translation.'   So with using the AIs, it is good to continue the
conversation with examples and variations on the theme.  Sometimes the AI
replies to me, "Yes, Julien, that is a good example."  That way I am more
confident that we are not talking apples and oranges.  (I must say, though,
that the AIs are too polite and affirming. No AI has ever replied to me,
"No, Julien, you are an idiot.")

Cheers,
Julien.

On Wed, Feb 28, 2024 at 11:50 AM Julien Winter <winter.julien at gmail.com>
wrote:

> Hi folks;
>
> Just for sport, I thought I would ask ChatGPD 3.5 and Perplexity a
> question about air flow:
>
> QUESTION: "How do the turbulent vs. laminar characteristics differ between
> downward flowing and upward flowing air?"
>
> Both AIs didn't give a good answer on the first try and required a
> follow-up clarification. Perplexity though I was asking for definitions of
> turbulent and laminar flow, and ChatGPT thought that downward-flowing air
> would bump into more obstacles.
>
> So, having sorted them out, this is what they said.
>
> ChatGPT ========================
>
> Certainly! When comparing the turbulent and laminar characteristics of
> downward flowing and upward flowing air without any obstacles obstructing
> the flow, the differences lie primarily in the effects of buoyancy and
> stability. Here's a revised comparison:
>
> Downward Flowing Air:
>
>
>    - Turbulence: Even without obstacles, downward flowing air can still
>    exhibit turbulence, especially if it encounters changes in temperature,
>    pressure, or wind shear. As the air descends, it may encounter different
>    atmospheric layers with varying properties, leading to turbulence.
>    - Laminar Flow: In ideal conditions with minimal atmospheric
>    disturbances, downward flowing air can exhibit more laminar flow compared
>    to upward flowing air. However, the presence of any instabilities or
>    variations in the atmospheric properties can disrupt laminar flow.
>
>
> Upward Flowing Air:
>
>
>    - Turbulence: Upward flowing air tends to be less turbulent compared
>    to downward flowing air, especially in the absence of obstacles. The
>    buoyancy force driving the air upward can result in a smoother, more
>    coherent flow pattern. However, localized turbulence can still develop due
>    to convective processes or changes in atmospheric conditions.
>    - Laminar Flow: Upward flowing air has a greater tendency to maintain
>    laminar flow in the absence of obstacles and under stable atmospheric
>    conditions. Buoyancy-driven flow, such as in thermals or updrafts, can
>    promote more organized, laminar flow patterns.
>
>
> In summary, in the absence of obstacles, downward flowing air can still
> exhibit turbulence due to the potential for encountering atmospheric
> disturbances, while upward flowing air tends to be smoother and more
> conducive to laminar flow, driven by buoyancy effects. However, both
> directions of flow can experience variations depending on the stability and
> properties of the surrounding atmosphere.
>
>
> Perplexity =====================================
>
> When comparing upward and downward flows of air in terms of turbulent and
> laminar characteristics, the differences lie in the behavior of the airflow:
>
> Upward Flow:
>
>
>    - Turbulent Characteristics: Upward turbulent airflow involves
>    irregular and chaotic movement with unpredictable fluctuations, mixing of
>    fluid particles, swirling vortices, rapid changes in velocity, and pressure
>    fluctuations.
>    - Application: Commonly observed in scenarios like smoke coming out of
>    exhaust pipes or turbulence in aircraft, upward turbulent flow is
>    influenced by irregularities or obstructions in the airflow path.
>
>
> Downward Flow:
>
>
>    - Laminar Characteristics: Downward laminar airflow is characterized
>    by smooth and streamlined movement where fluid layers move parallel to each
>    other without mixing or disruption.
>    - Observation: In downward laminar flow, the fluid flows in a straight
>    line with low flow rate and constant properties like velocity and pressure.
>
>
> In essence, upward flowing air tends to exhibit turbulent characteristics
> with chaotic movement and mixing of particles, while downward flowing air
> typically displays laminar behavior with smooth and orderly movement of
> fluid layers. These distinctions are crucial for various applications where
> understanding airflow dynamics is essential for system design and
> efficiency.
>
> ==================================
>
> With some iterative, leading or loaded questions, it might be possible to
> push the AIs into the answer you want.  But that is a human training the
> AI, not the AI training the human.
>
> Niether AI seems to have read the book Norbert suggested: Groume-Grjikailo
> in his 1923 book "The Flow of Gases in Furnaces"  on pages 88 - 91.
> https://archive.org/details/TheFlowOfGasesInFurnaces/page/n107/mode/2up?view=theater
>
>
> It is also interesting to see how it is good to use several AIs, because
> they give different answers.
>
> Have a whack at it and see what you get.
>
> Cheers,
> Julien.
>
> --
> Julien Winter
> Cobourg, ON, CANADA
>

-- 
Julien Winter
Cobourg, ON, CANADA
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.bioenergylists.org/pipermail/stoves_lists.bioenergylists.org/attachments/20240228/e6ec3516/attachment.html>