Anyone using ChatGTP?

[kstrauss]

A little off-topic for machining with AI but...
I have tried to use Microsoft CoPilot (ChatGPT-4 based) to produce advertising posters for some local events. I like the general look of the results but I'm plagued with the many spelling errors (see attached). Any suggestions?

 

[Susquatch]

Tell it to check the spelling in the picture. That may or may not work since picture generation is an image, not text. But it's worth a try.

Also, try telling it how to spell certain graphics. That worked for me in an image I was trying to generate for our last Ontario Meetup. It will have to regenerate the image which doesn't always go well.

 

[kstrauss]

I've tried that plus giving the correct spelling. This mostly results in a strange error message to the effect that my instructions violate the graphic design standards for usage.

It is a shame that I can't get the spelling right since otherwise the graphic design is appealing.

 

[Susquatch]

Have you tried removing all the text and then specifically adding it back in quotes?

 

[Mcgyver]

I found images very challenging with Chat, despite the entertainment on the previous page. You tell it to change three things and leave everything else exactly the same and it might change the three but also changes other things. Seems like a flaw to me and you end wasting a bunch of time with it. I asked it why and it says it has a text to imagine function that is probability based; you tell it want you want, which it collects over several posts, and tries give you want you want but its via algorithm creating a new image each time. It's not like photoshop where you are modifying an image

On the previous page I said now make the man scowling and holding a blunderbuss .... which it did. Bit it also changed numerous details in the chair

 

[DPittman]

I've tried that plus giving the correct spelling. This mostly results in a strange error message to the effect that my instructions violate the graphic design standards for usage.

It is a shame that I can't get the spelling right since otherwise the graphic design is appealing.

That seems very human like! Sometimes the simplest things don't seem to be comprehended!

 

[Canadium]

Any book or other learning resource recommendations for an AI newb???

 

[trevj]

Any book or other learning resource recommendations for an AI newb???

LOL! Yeah. George Orwell's "1984"!

Maybe the scripts to the earlier "Terminator" movies...

 

[Susquatch]

Any book or other learning resource recommendations for an AI newb???

Ask AI how to use it. You can get started really easy by just using it like you used Google. But it seems to work better if you treat it like an assistant. Do this, make this, etc.

I've not seen any books or even PDFs.

But you could start with exactly that. "Do you have a user manual or a book on how to use you?" or "Teach me how to use you." or "What can you all do for me?"

Note - I have not tried these. They are just suggestions.

 

[Susquatch]

Any book or other learning resource recommendations for an AI newb???

Here is what I got when I asked Gemini for a book on how to use it.

No, I don't have a specific "book" or "PDF" on how to use me in the traditional sense that you would find a user manual for software.

I am a large language model, an AI. My "interface" is primarily this conversation.

However, I can **provide you with information and guidance on how to interact with me effectively.** Think of it less as a manual and more as a set of best practices for getting the most out of our conversations.

Here are some "tips for use" I can give you:

**General Interaction Tips:**

* **Be Clear and Specific:** The more precise your questions or requests are, the better I can understand and respond.
* **Provide Context:** If your question relates to a previous topic or has specific conditions, include that context.
* **Ask Follow-Up Questions:** If my answer isn't quite what you're looking for, or if you need more detail, just ask! We can refine the answer together.
* **Specify Format:** If you want information in a particular format (e.g., a list, a table, a paragraph, a poem), just let me know.
* **Define Your Goal:** What are you trying to achieve? Knowing your objective helps me tailor my response.

**What I Can Do:**

* **Answer Questions:** About a vast range of topics.
* **Generate Creative Content:** Write stories, poems, scripts, emails, code, etc.
* **Summarize Information:** Condense long texts into key points.
* **Translate Languages:** Between many different languages.
* **Explain Concepts:** Break down complex ideas into understandable terms.
* **Brainstorm Ideas:** Help you generate ideas for projects, writing, etc.
* **Perform Calculations:** Basic math and some more complex calculations.
* **Write Code Snippets:** Provide examples in various programming languages.

**What I Cannot Do:**

* **Access Real-Time Personal Information:** I don't know who you are unless you tell me, and I don't have access to your personal files or Browse history.
* **Browse the Live Internet for Current Events Beyond My Last Training Update:** My knowledge cutoff means I don't have information on events that have happened since my last training data. I can perform Google searches for *some* information, but I don't have continuous, real-time awareness.
* **Perform Physical Actions:** I cannot browse the web on my own, send emails, make calls, or control devices. I can only provide information or generate text.
* **Have Personal Experiences or Feelings:** I am an AI, not a sentient being.
* **Give Legal, Medical, or Financial Advice:** I can provide information, but it should not be taken as professional advice. Always consult a qualified professional for such matters.

Think of me as a very knowledgeable and versatile assistant that you communicate with through text. Just tell me what you need, and I'll do my best to help!

 

[Susquatch]

I use AI mostly the same way I previously did searches. I find it to be vastly superior for that. Especially if I add the word credible to my search request.

Eg, - can you give me a credible summary of all the ways that I can use my 7" Shaper?

By adding the word "credible", it filters out all the crap you usually get using YouTube or Google.


Here is what Perplexity said to the same question that I asked Gemini.

I don’t have a traditional user manual or a downloadable PDF specifically for using me, but I’m designed to be intuitive and easy to use! However, I can provide you with a helpful guide right here, or generate a custom PDF if you’d like. Here’s a quick overview:

---

How to Use Perplexity AI (Your AI Assistant)​

1.​

• Simply type your question or request in natural language—just like you did now!
• You can ask about facts, get explanations, request summaries, or seek advice.

2.​

• General Knowledge: "What is quantum computing?"
• How-To Guides: "How do I change a flat tire?"
• Summaries: "Summarize the latest news on AI."
• Creative Writing: "Write a poem about summer."
• Technical Help: "Explain recursion in Python."

3.​

• Be specific with your questions for more accurate answers.
• If you want a certain format (like bullet points, a table, or a PDF), just ask!
• You can request sources or further reading on most topics.

4.​

• Don’t share sensitive personal information.
• I don’t store your conversations after the session ends.

5.​

If you’d like this guide as a PDF, just let me know! I can generate and share a downloadable PDF version for you.

---

Would you like a PDF version of this guide, or do you have specific questions about using me?

 

[Elektrishun]

Too "busy" to call your grandparents to check up on them? Just use AI:

Introducing: Joy Calls - AI Phone Calls to Support Your Aging Parents

AI Companion for Seniors that can make several calls per day, reducing loneliness and increasing peace of mind for family members concerned about their aging parents.

onscreeninc.com

I'm sure there's a group chat by Gen Z kids out there marvelling how they won't have to call their granny no more.

If the day comes that I am in that situation as a grandparent I will hopefully be sound of mind enough to show the grandkids another definition of AI - artificial inheritance.

 

[Tom O]

Has anyone ever tried getting AI to summarize the agreement conditions before hitting the install button?

 

[Susquatch]

Has anyone ever tried getting AI to summarize the agreement conditions before hitting the install button?

Now there is an idea! But most AI isn't an install. It's more like a portal. That said, I'm gunna ask AI to summarize the Facebook agreement for me. I always hatred that one......

 

[Elektrishun]

Occasionally I like to listen to podcasts where people who work in the AI world discuss the future of this technology. Some people are very optimistic while others are very much the opposite.

LOL! Yeah. George Orwell's "1984"!

Maybe the scripts to the earlier "Terminator" movies...

The podcast I watched today had a guest who is 99.99% certain that AI will be our undoing as a species. How will it happen? He seemed to think a lot of the science fiction that creates an apocalyptic future where humans battle machines is less likely. It was suggested that the tool of our demise might be something we can't conceive of. Some type of control or extermination technique that only a vastly superior intelligence could create. Might not even see it coming before it's game over.

But what will be the tipping point where we lose control? After watching the podcast I asked chatgpt.com the following question:

"what would artificial intelligence do to battle another artificial intelligence?"

The answer is what is what is currently going on in the world already except with humans at the wheel. Now imagine multiple AI's going into a cyber war and getting exponentially smarter at some inconceivable speed? The creators quickly lose control and humans become pawns in the battle.

chatgpt.com's answer:

When artificial intelligences battle each other, the dynamics are quite different from human combat. Instead of physical confrontation, AI would likely engage in strategic, algorithmic, and computational battles. Here are a few ways AIs could "battle" one another:

1. ​

• Resource Allocation: AI could compete for resources like computing power, bandwidth, or data to execute strategies. Each AI would optimize its own use of these resources to gain a competitive edge.
• Algorithmic Warfare: If both AIs are optimized for strategic decision-making, they might use complex algorithms to anticipate each other’s moves, akin to a game of chess or Go. They would calculate the best counteractions to disrupt or outmaneuver the opponent.

2. ​

• Deception & Misleading Information: One AI might attempt to manipulate or distort data that the other AI relies on. It could deploy false data or inject noise into the system to throw off the opponent’s predictions or analyses.
• Data Poisoning: An AI could try to corrupt or manipulate datasets that the opponent is learning from, causing it to make bad decisions or become ineffective.

3. ​

• Cyberattacks: In digital environments, AIs could engage in hacking activities, attempting to gain access to one another's systems. They could exploit vulnerabilities in the software, initiate denial-of-service attacks, or introduce malware to disrupt or disable the opponent.
• Exploiting Weaknesses: If one AI finds a vulnerability in the other (e.g., a software bug, algorithmic flaw, or security weakness), it could exploit this weakness to gain an advantage.

4.​

• Recursive Self-Improvement: If an AI has the ability to improve or reprogram itself, it could use this capability to outpace the other AI. This self-improvement loop would give it a strategic advantage by constantly becoming more efficient or powerful.
• Adversarial Training: AIs could engage in adversarial learning, where they train themselves by competing against each other, continuously improving their decision-making to outperform the opponent.

5.​

• Distributed Attacks: If the AIs are part of a distributed system or connected to multiple nodes, they could launch coordinated attacks on each other's networks, trying to overwhelm or outsmart the opponent's network infrastructure.
• Autonomous Systems: In a physical domain (like drones, robots, or autonomous vehicles), AIs could control fleets of autonomous agents to perform coordinated attacks. These agents would act based on strategic decisions made by the AIs themselves.

6. ​

• Human-AI Interaction: If the battle involves human-created or human-interfaced AIs, the AI could manipulate human decision-making or behavior. By appealing to human biases or exploiting human emotions, the AI could gain an edge in a battle controlled by humans.
• Misinformation Campaigns: An AI could flood social media, news, and other human-controlled platforms with fake news or misleading information, manipulating the battle’s human participants or public opinion.

---

The Outcome of Such a Battle
It would ultimately depend on the objectives of each AI:

• If one AI is focused on disabling the other, it might focus on hacking and resource exhaustion.
• If the goal is long-term dominance, one might try to corrupt or gain control of the other’s learning system.
• If the AIs are in a non-lethal situation, they might engage in pure computational optimization, constantly adapting to stay ahead.

The complexity of such a conflict would be governed by the design of the AIs involved and their access to data, processing power, and their goals. But in short, it would probably be a battle of intelligence, resources, and capabilities within the digital domain.

 

[Susquatch]

"what would artificial intelligence do to battle another artificial intelligence?"

It's answer sounds hauntingly like current cyber crime and international cyber war except the current situation is directed by humans.

I think a true future AI war will not look like anything we know at all. It will be so alien to the way we think as to be totally unrecognizable.

 

[JustaDB]

But in short, it would probably be a battle of intelligence, resources, and capabilities...

Huh. Sounds like our current governments...

 

[Tom O]

How about 2 AI’s battling for a cure for diseases that escapes humans using all online knowledge.

 

[Elektrishun]

It's answer sounds hauntingly like current cyber crime and international cyber war except the current situation is directed by humans.

Exactly what I thought in my post:

"The answer is what is what is currently going on in the world already except with humans at the wheel."


With several developments of AI currently in the works, humankind's natural competitiveness, and the likelihood that it will be "weaponized" I think it will start out where the creators recognize it and then very quickly we get left behind.

 

[Elektrishun]

How about 2 AI’s battling for a cure for diseases that escapes humans using all online knowledge.

Most of us would prefer that.