ChatGPT incoming

ChatGPT was released to the public a few weeks ago...it is not yet (allegedly) connected to the WWW.

It can answer questions, write code, write thesis, do graphic art etc.

I don't have a software or hardware background so Dave Jones explains it better...

'ChatGPT & AI Has Changed EVERYTHING' - https://www.youtube.com/watch?v=cHyyXSEXfus

Ches

PS: Wiki summary ... ChatGPT - https://en.wikipedia.org/wiki/ChatGPT

Looks like it's up and running, but you need an account to try it.

Maybe not, dreadful to find that an Artificial Intelligence knew more about Model Engineering than me. It's not a tall order, I set a low bar...

It's early days yet, but I have a feeling this might not be good news for a lot of professions. On the other hand lawyers, say, could use it to prepare the reams of paperwork required for a court case.

Apparently, if asked, it can instantly write Arduino and Python code if given an objective.

Elon has constantly said that AI is potentially the biggest threat to humanity. He has been involved in the Open AI company that is currently 'containing' ChatGPT.

Ah well, 2023 could be interesting :-#

Ches.

Well, I'm not too worried yet. The hype usually outstrips the reality, except for carefully crafted demonstration pieces.

Rob

As I understand it, the model is incapable of expressing uncertainty, and is compelled to write whatever it can on a given subject even when it doesn't have good data on which to base it's output.

This has arisen because it was trained on a "corpus" of information which was predominantly highly authoritative sources... And is at least in part a "black box" which can't be unpicked and adjusted with ease.

.

The net result is like combining tha bloke down the pub who is "always" right with a speech-writer who always manages to make politicians sound earnest and believable; doesn't seem like a sensible thing to trust.

.

It also took less than 2 weeks for a uni student in the States to write some code which can accurately detect when GPT has been used to write something, so it must actually be quite formulaic when you look at a large number of examples.

a uni student in the States to write some code which can accurately detect when GPT has been used to write something,

Yes, I saw that. But the ChatGPT has only been alive for a few weeks. I wonder how refined it will become in the next 12 months...especially if they give it it's own unlimited internet connection?

Just imagine how convincing Spam phone calls and email phishing will become?

Where's Arnie when you need him

Ches

What we need is Artificers Intelligence.

🥴

Artificial intelligence will never be a match for human stupidity.

10 Jan 2023 - Microsoft about to buy 49% of OpenAI (+ChatGPT) .... https://www.cnbc.com/2023/01/10/microsoft-to-invest-10-billion-in-chatgpt-creator-openai-report-says.html

Microsoft will reportedly get a 75% share of OpenAI’s profits until it makes back the money on its investment, after which the company would assume a 49% stake in OpenAI.

As usual, with anything MS tends to interfere with, this will probably not end up well for anyone. Seems they have plans to incorporate it in to Word, Excel etc and whatever else.

Ches

First off I don't know exactly if this crosses a line so consider this a tester.

If this annoys Moderators or other members then please say so and I will not mention it again.

Here are the some results of some chats.

Q. What can you generalise about this website https://www.model-engineer.co.uk/forums/

A. Thank you for providing the website URL. Based on the website **LINK**, I can estimate the following posting statistics for the forum:

1. There are currently over 30,000 registered members on the forum, based on the "Members" page of the website.

2. The forum has over 1.3 million posts, based on the "Forum" page of the website. This suggests that the forum is a well-established community with a significant amount of user-generated content.

3. New posts are added to the forum regularly, with some topics having multiple posts added per day. The frequency of new posts can vary depending on the popularity of the topic being discussed.

4. The forum has a wide range of topics related to model engineering, including discussions about tools and machinery, steam engines, and model railways.

It is important to note that these estimates are based on publicly available information and may not be completely accurate. Additionally, the posting statistics may vary over time depending on factors such as user engagement levels and the popularity of different topics.

Q. what can you generalise about this website https://www.model-engineer.co.uk/forums/latest_posts.asp

A. Based on the website **LINK**, I can make the following general observations:

1. The website appears to be a forum focused on discussions related to model engineering, with various topics being discussed by members of the forum.

2. The website design is fairly basic, with a predominantly white color scheme and simple layout. The website seems to be designed more for functionality than aesthetics.

3. The website has a clear and straightforward navigation menu, making it easy for users to find the information they are looking for.

4. The website appears to have an active user base, with new posts being added to the forum regularly.

5. The website seems to be primarily aimed at model engineering enthusiasts, and may not have as broad an appeal to a more general audience.

6. The website appears to be relatively well-established, with a history dating back to 1898 according to the website's "About Us" page.

Overall, the website appears to be a well-established forum catering to a specific niche audience of model engineering enthusiasts. While the design may be basic, it seems to serve its purpose effectively and the active user base suggests that the forum is a valued resource for those interested in model engineering.

Q. And ditto for the search function

Thank you for providing the search page of the website **LINK**.

Unfortunately, the search page appears to be a general search function for the entire website, and it does not specifically search the forum.

To search for conversations about tensile structures on the forum, I recommend going to the forum section of the website and using the search function there.

Here's how to do it:

1. Go to the forum section of the website: **LINK**
2. Click on the "Search" button located on the right-hand side of the page.
3. In the search bar that appears, type in "tensile structures" (without quotes) or any relevant keywords related to the topic you want to search for.
4. Click the "Search" button to begin your search.

The search results will display any relevant conversations related to your search terms. If you have any trouble with the search function or need further assistance, please let me know.

Make off it what you will.

Based on my limited interactions - This is significantly different - I'm all in.

You may see things differently.

Regards.

Frank C.

Wow, I'm impressed

That's not really a chat, though, is it? Asking a bot to regurgitate what it's sifted from trawling a website is more or less asking a bot to do the one thing it was primarily "born" to do.

On the other hand, I would be very interested to see an answer from ChatGPT to the essay question:

"In Flaubert's Madame Bovary Charles is largely to blame for Emma's death". Discuss. Refer closely to the text in your answer."

Like driverless cars .... remember those?

Well we knew Percival Marshall was ahead of the game...... so he must have invented the Internet.

Come on, you've got to ask it the most common question: What lathe should I get?

The problem with ChatGPT is that it can talk, it just doesn't know anything else. In a sense it's like a incredible lossy compression of the internet. It can reproduce something that looks like a "reasonable internet reply" when it comes to language, vocabulary, sentence structure and so on. But it doesn't actually know anything so the answer can often be both perfectly believable and positively wrong.

Combining ChatGPT with Wolfram Alpha sounds like a good idea, but I also think we are jumping to conclusions based on very simple "tricks". It's not intelligence, just very clever pattern matching.

I asked it 'Is the Kugelpanzer' a hoax, and it wrote a well argued case that it is a hoax, but offered some evidence on both sides.

I asked Bard, and it was adamant that the Kugelpanzer is real.

(It's a good question as there is a 'Kugelpanzer' in a Russian tank museum, but it's authenticity and provenance are disputed, so a nuanced subject).

N.

I asked it to choose between two options:

"Both the Myford Super 7 and the SC4 lathe are high-quality machines that are suitable for advanced hobbyists and professionals. The decision on which one to choose ultimately comes down to your individual needs and preferences.

Don’t even think about getting it to write articles for MEW.

regards Martin

.

I don’t want to get involved with the wretched thing, but could you please see how well it can describe the distinction betweem precision and accuracy ?

Most of us would likely resort to graphics of targets, so I would be intrigued to read its verbal description.

MichaelG.

Here you are Michael,

Precision and accuracy are both terms that are often used in the context of measurement and data analysis, and they refer to different aspects of how well a measurement or calculation represents the true value or intended result.

Accuracy refers to how close a measurement or calculation is to the true or intended value. In other words, it measures how well a measurement reflects reality or how correct a calculation is. For example, if you measure the length of a table with a ruler and get a result of 1.02 meters, but the actual length is 1 meter, your measurement is not accurate.

Precision, on the other hand, refers to how consistent or reproducible a measurement or calculation is. In other words, it measures how well a measurement can be repeated and how consistent the results are. For example, if you measure the length of the same table multiple times with the same ruler and get results of 1.02 meters, 1.03 meters, and 1.01 meters, your measurements are precise but not necessarily accurate.

In summary, accuracy measures how close a measurement or calculation is to the true or intended value, while precision measures how consistent or reproducible a measurement or calculation is.