I asked ChatGPT to do a 500 word paper about nixie tubes...

[Jeff Walton]

Recently, there has been news coverage about an AI tool named ChatGPT.  It can supposedly scrape information from the internet and generate a "human" response to questions. 

I tried asking,   "Create a 500 word essay about nixie tubes" and here is what it came up with:

Nixie tubes, also known as cold cathode displays, were a popular type of electronic display used in the 1950s and 1960s. The name "Nixie" is a trademark of Burroughs Corporation, which was the main manufacturer of these devices. Nixie tubes consist of a sealed glass or plastic envelope that contains a wire mesh anode and multiple cathodes in the shape of numerals or other symbols. When a high voltage is applied to a cathode, it emits a bright orange-red glow, allowing the tube to display numbers or other information.

One of the main advantages of nixie tubes is their high visibility, even in low light conditions. They are also relatively durable and long-lasting, with a typical lifespan of around 10,000 hours. Additionally, nixie tubes have a unique retro aesthetic that has led to a resurgence of interest in them among hobbyists and designers.

However, there are also several disadvantages to using nixie tubes. They require a high voltage to operate, typically around 170 volts, which can be dangerous if not handled properly. They also consume a significant amount of power, which can be a drawback in battery-powered applications. Additionally, nixie tubes are relatively large and bulky, which can be a challenge in designing compact electronic devices.

Despite these limitations, nixie tubes were widely used in a variety of applications, such as digital clocks, voltmeters, frequency counters, and other types of test equipment. They were also used in some early computers, such as the UNIVAC I and the IBM 702. In the 1960s, nixie tubes began to be replaced by other types of electronic displays, such as vacuum fluorescent displays (VFDs) and light-emitting diodes (LEDs), which offered lower power consumption and higher reliability.

In recent years, there has been a renewed interest in nixie tubes among hobbyists and designers. Many people have begun building their own nixie clock projects, using vintage or new-old-stock (NOS) tubes, as well as more modern designs that incorporate microcontrollers and other modern electronic components. Additionally, some companies have started to produce new nixie tube designs, such as the IN-14 and Z573M tubes, which are compatible with modern electronics and have a longer lifespan.

Another interesting application of nixie tubes is using them as an artistic medium. Many artists have used nixie tubes in sculpture, installation and other forms of art, mostly because of their distinct and unique aesthetic. Nixie tube art has been featured in galleries and exhibitions all over the world, and many of these pieces are highly sought after by collectors.

In conclusion, Nixie tubes were an important technology in the early days of electronic displays, and they continue to be popular among hobbyists and designers today. Their high visibility, durability, and retro aesthetic make them an attractive option for a wide range of applications. However, the high voltage required to operate nixie tubes, as well as their power consumption and bulky size, make them less suitable for some modern electronic devices. Despite this, the unique aesthetic of Nixie Tubes continues to inspire and be used in a variety of projects and applications, including the arts.

[Mac Doktor]

On Jan 21, 2023, at 12:52 PM, Jeff Walton <jwalt...@gmail.com> wrote:

I tried asking,   "Create a 500 word essay about nixie tubes" and here is what it came up with:

It's close, 513 words.

https://openai.com/blog/chatgpt/

The site says it's at capacity right now, try again later.

Terry Bowman, KA4HJH
"The Mac Doctor"

https://www.astarcloseup.com

“...the book said something astonishing, a very big thought. The stars, it said, were suns but very far away. The Sun was a star but close up.”—Carl Sagan, "The Backbone Of Night", Cosmos, 1980

[Toby Thain]

On 2023-01-21 12:52 p.m., Jeff Walton wrote:
> Recently, there has been news coverage about an AI tool named ChatGPT.
> It can supposedly scrape information from the internet and generate a
> "human" response to questions.
>

> I tried asking, *"Create a 500 word essay about nixie tubes"* and here

> is what it came up with:
>

AI scams are off topic.


> --
> You received this message because you are subscribed to the Google
> Groups "neonixie-l" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to neonixie-l+...@googlegroups.com
> <mailto:neonixie-l+...@googlegroups.com>.
> To view this discussion on the web, visit
> https://groups.google.com/d/msgid/neonixie-l/877b5451-9be7-4cc0-a2b6-505fff0a070bn%40googlegroups.com <https://groups.google.com/d/msgid/neonixie-l/877b5451-9be7-4cc0-a2b6-505fff0a070bn%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Nicholas Stock]

That doesn't make sense Toby. Why is this a scam?

To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/35ca88c2-2eaa-10b1-40ba-608933361ee1%40telegraphics.net.

[Dekatron42]

There are some interesting discussion over at Nature.com regarding scientific articles that have ChatGPT listed as an author. But what also caught my eye while reading those articles is that it exists websites who offers authorship as services, just pay them enough and they'll make sure that a famous (probably not respectable though) scientist will co author your paper an put his name on it....

ChatGPT has been discussed when it comes to references that it generates and what I've read so far is that all references were false, it didn't even reference the articles it leached its texts from.

I'm glad we don't have an AI programming code generator that works this badly, it is enough with all poor programmers out there..... ;)

/Martin

[John DeArmond]

This thing is amazing and maybe a little scary. I tried to stump it
with "explain deflagration transition to detonation". (OK, so I'm an
explosives expert too :-)

Response:

"Deflagration is a type of combustion in which a flame front propagates
through a substance at a subsonic velocity, releasing energy in the form
of heat and pressure. Detonation, on the other hand, is a supersonic
combustion process in which a shock wave precedes the flame front,
causing a rapid release of energy.

The transition from deflagration to detonation, also known as
deflagration-to-detonation transition (DDT), occurs when the pressure
and temperature generated by the deflagration process reach a certain
threshold, causing the flame front to accelerate and become supersonic.
This transition is often associated with a significant increase in the
rate of pressure rise and release of energy, making it a potentially
dangerous phenomenon in certain situations, such as in the combustion of
fuels and explosives.

DDT is a complex process and its understanding is still not complete.
Many factors can influence the transition, including the properties of
the combustible mixture, the geometry of the combustion chamber, and the
initial conditions of the combustion process.

In order to prevent or control DDT, various methods have been developed,
such as the use of flame arresters, which are devices that inhibit the
spread of a flame front by disrupting the chemical reactions that
sustain combustion. Others methods include the use of specific ignition
systems, or the use of additives that change the reaction kinetics of
the fuels."

Wow! Spot on. Exactly correct.

I then asked it "Explain nixie tube poisoning". It didn't do too well
on that one.

John

On 1/21/23 13:34, Mac Doktor wrote:
>
>> On Jan 21, 2023, at 12:52 PM, Jeff Walton <jwalt...@gmail.com

>> <mailto:jwalt...@gmail.com>> wrote:
>>
>> I tried asking, *"Create a 500 word essay about nixie tubes"* and here

>> is what it came up with:
>
> It's close, 513 words.
>
>

> https://openai.com/blog/chatgpt/ <https://openai.com/blog/chatgpt/>

>
> The site says it's at capacity right now, try again later.
>
>
> Terry Bowman, KA4HJH
> "The Mac Doctor"
>

> https://www.astarcloseup.com <https://www.astarcloseup.com>

>
> “...the book said something astonishing, a very big thought. The stars,
> it said, were suns but very far away. The Sun was a star but close

> up.”—Carl Sagan, "The Backbone Of Night",/Cosmos/, 1980

>
>
> --
> You received this message because you are subscribed to the Google
> Groups "neonixie-l" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to neonixie-l+...@googlegroups.com
> <mailto:neonixie-l+...@googlegroups.com>.
> To view this discussion on the web, visit

> https://groups.google.com/d/msgid/neonixie-l/8468AF9D-6637-42B7-ADCB-4351D2BEC48C%40gmail.com <https://groups.google.com/d/msgid/neonixie-l/8468AF9D-6637-42B7-ADCB-4351D2BEC48C%40gmail.com?utm_medium=email&utm_source=footer>.

--
John DeArmond
j...@neon-john.com
j...@johndearmond.com

[gregebert]

John - How long did it take for 'HAL' to compose the response ? That would give a lot of clues about how far-and-wide it searches for info. I'm guessing it took at least several minutes.

[Audrey]

It doesnt search for info to generate responses. This model could be run entirely offline.

To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/50ad5252-f240-4152-b35e-66b8697132ban%40googlegroups.com.

[Will Yager]

All of GPT-3's world-knowledge is embedded into its parameters. Whatever it tells you is the equivalent of a human telling you something "from memory". It takes only a few seconds for it to respond to a prompt.

Connecting AIs to external data sources is, I imagine, under active research, but none of the current LLMs use external data.

> On Jan 24, 2023, at 13:24, gregebert <greg...@hotmail.com> wrote:
>
> John - How long did it take for 'HAL' to compose the response ? That would give a lot of clues about how far-and-wide it searches for info. I'm guessing it took at least several minutes.
>

[gregebert]

I'm having a hard time grasping how a system could store and index so much information locally. That in itself is daunting due to the sheer volume of data. The other thing I'm curious about is how much of the response is plagiarized/reworded from existing information, versus true AI (collated from various sources, analyzed for consistency, and then cherry-picked for the best info without being redundant or missing important tidbits).

[Adrian Godwin]

The examples I've read 'feel' very plagiarised. They read like those web pages put out there to attract google hits and, thereby, accidental advertising views - very much as thoughj ripped from wikipedia or advertising copy. At best, I would think they pull a bunch of paragraphs from different sources.

There is apparently research happening to analyses these articles (with more 'AI' !) to try to catch undergrad essays created using the bots. It will be interesting to try those on those search engine targets .. maybe google will then demote them in favour of original material.

--

You received this message because you are subscribed to the Google Groups "neonixie-l" group.

To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/360c505b-7b50-4230-879d-0a31b1bf6b85n%40googlegroups.com.

[Audrey]

Yeah it is a lot to wrap your head around, the data isn't really stored per se, it's trained on the text word-by-word and it tries to guess what word comes next after the previous word, and that shapes its parameters, the text itself is never stored per se, it just tries to answer your queries based on what it thinks the response is, the training data can't cover every possible niche topic because they havent fed it every bit of text there is, that's probably why it didnt understand a question around cathode poisoning. In some cases it does spit out articles largely verbatim, but that's undesired and unintended, and usually occurs when there isnt a lot of data on the subject (called overfitting).

--

You received this message because you are subscribed to the Google Groups "neonixie-l" group.

To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/360c505b-7b50-4230-879d-0a31b1bf6b85n%40googlegroups.com.

[Nicholas Stock]

As others have described, any AI's output is very dependent on the training data provided but also the nature of the neural network that is used for that training (garbage in, garbage out still holds for any machine learning process). For some staggeringly successful examples, check out AlphaGo (https://www.nature.com/articles/nature16961) and Alpha Zero (https://www.deepmind.com/research/highlighted-research/alphago). This technology lies behind image recognition software, natural language processing and is slowly making its way into multiple different areas of society (I was head of chemistry at an AI based chemistry startup for 3 years, so I looked at this quite a bit...). Now ChatGPT..... who knows what's next!

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/CAPDJQ6Cm-c6%2Bs0L8cxqOqbABSzd0LMWG3yGut2DnS5UkM9sd%3Dg%40mail.gmail.com.

[Jeff Walton]

This discussion is the reason for my original post.  Some of the details regarding nixies and reliability, etc appeared to come from sites that most on the group are familiar with.  It must be using web based information to come up with some of the specifics.  The interpretation is the interesting part.  No general database is going to know about direct drive VS multiplex and such with any detail. 

[newxito]

I think we're going to see hundreds of copyright lawsuits against these AI companies and their users. It started already with Copilot...