Anyone using ChatGTP?

[trevj]

Most of us would prefer that.

But I cannot, as far as I see it, see any version of AI, thinking that people are anything but a problem to be solved...

 

[Susquatch]

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

I understand that is actually happening with some degree of success. But then, we have the approval process to delay implementation.

Here is a parallel thought. Imagine the good that could be done by reallocating climate change funding to other global problems like hunger and disease.

 

[Gennady]

Folks, don't need to look at AI as human being - it is not. It is just super powerful search engine, complier of data and collate engine. Nothing more. Try to ask questions about feeling, sensitivity, self awareness, conciseness, (God forbid - about love) you will see very dry prebuilt answers. Use at as a tool not as a friend or living soul.
I am using paid version of Claude and I have to say it is beyond amazing as technical tool or even counsellor. I am always tempting to say "Please..." when I am requesting to generate code for QEI interface for PIC18F4331 but than I am catching myself that I am talking to a silicone chip.

 

[Susquatch]

Folks, don't need to look at AI as human being - it is not. It is just super powerful search engine, complier of data and collate engine. Nothing more.

I certainly agree for those things that "WE" use AI for. I am not familiar with Claude. However, I had several debates with Gemini and Perplexity and I can say that they do an incredible job - WAAAY better than most humans do. Raw debate with no emotions. Amazing really.

I've also had several phone conversations with AI that was being used as a front end for customer service. We were talking for several minutes before I realized that something was wrong. It was that good. In fact, that is what tweaked me. It was too good!

I also know of several large corporate efforts to replace sales staff with AI. My son had the same last name as the head of a large corporation selling insurance. So he got a lot of his email. They are working on an AI front end to replace their sales staff lock stock and barrel. Imagine calling a broker for a quote. When you call, he validates you right then and there. He also knows your insurance and driving record, knows when you lie, and evaluates your personality to upsell you. 100,000 employees out of a job.

I also know that cyber crime has been working on AI that can fake your own voice after 2 or 3 sentences, then call your wife asking for your bank password or other nefarious purposes. She would NEVER KNOW it wasn't you! It wouldn't work on me cuz my wife doesn't know any of my passwords. But a lot of spouses do. And that's just an example.

Here is a summary of what we "know" the US Military is doing. Imagine what we don't know....

The U.S. military currently has a vast portfolio of AI projects, with over 800 active initiatives across defense branches. These projects focus on areas like process efficiency, threat evaluation, battlefield decision-making, autonomous systems, and intelligence analysis.

A flagship effort is Project Maven, launched in 2017, which uses machine learning and data fusion to process sensor data (satellite imagery, UAV video, etc.) to identify potential military targets. Maven’s system supports recognizing and prioritizing targets to enable quicker, more accurate strikes.

More recently, the Department of Defense established a Generative AI Task Force (Task Force Lima) in 2023 to coordinate generative AI adoption across defense operations, focusing on enhancing decision support, target recognition, casualty care, and reducing operator cognitive load. This represents a second phase of military AI evolution, where chatbot-like generative models aid intelligence analysis and provide interactive support for commanders, improving situational awareness and operational speed.

I'd be totally shocked to learn that AI isn't being developed for espionage and detailed analysis of large model communication systems like txt messaging, Facebook, etc. A human could never do this but AI could sift through that info in a heart beat and instantly highlight conversations like this thread sifting for new ideas and bad actors, etc

Imagine an AI hacker going after you. It profiles you, builds a complete record of everything you have ever said on line, or published in the newspaper or was said by others about you. It learns your personality quirks and your weaknesses. It does all that in a few minutes and then it deliberately targets you.

And how about terrorist initiatives looking for the best way to kill the most people. That one makes me lose sleep.

For now, I like AI though. It does things for me faster than I could do them myself. I especially like a good non-emotional debate.

 

[Elektrishun]

Folks, don't need to look at AI as human being - it is not. It is just super powerful search engine, complier of data and collate engine. Nothing more.

Hmmm...
That's exactly what an AI would want us to think before taking over the world.

 

[Elektrishun]

I certainly agree for those things that "WE" use AI for. I am not familiar with Claude. However, I had several debates with Gemini and Perplexity

• Google Gemini: Google's AI model, known for high-level reasoning, coding, and integration with Google products.
• Microsoft Copilot (formerly Bing Chat): Microsoft's AI, deeply integrated into Windows and Microsoft Office, offering productivity enhancements.
• Claude: Developed by Anthropic, it is praised for human-like writing and a strong emphasis on user privacy regarding data usage for model training.
• Perplexity AI: Stands out for its strong online research capabilities, providing accessible and verifiable answers.
• Jasper.ai: A popular choice for enterprise content creation and marketing copy.
• DeepSeek AI: A Chinese-made AI model noted for being cost-effective and a direct competitor in the AI market.
• HuggingChat: An open-source alternative for those interested in exploring AI projects.
• Pi (Inflection AI): Focuses on providing empathetic and supportive conversations.
• GitHub Copilot: Primarily focused on assisting developers with coding.
• Chatsonic: Known for its voice chat capabilities.
• Writesonic: A platform for AI-powered copywriting and content creation.

 

[Susquatch]

• Google Gemini: Google's AI model, known for high-level reasoning, coding, and integration with Google products.
• Microsoft Copilot (formerly Bing Chat): Microsoft's AI, deeply integrated into Windows and Microsoft Office, offering productivity enhancements.
• Claude: Developed by Anthropic, it is praised for human-like writing and a strong emphasis on user privacy regarding data usage for model training.
• Perplexity AI: Stands out for its strong online research capabilities, providing accessible and verifiable answers.
• Jasper.ai: A popular choice for enterprise content creation and marketing copy.
• DeepSeek AI: A Chinese-made AI model noted for being cost-effective and a direct competitor in the AI market.
• HuggingChat: An open-source alternative for those interested in exploring AI projects.
• Pi (Inflection AI): Focuses on providing empathetic and supportive conversations.
• GitHub Copilot: Primarily focused on assisting developers with coding.
• Chatsonic: Known for its voice chat capabilities.
• Writesonic: A platform for AI-powered copywriting and content creation.

Good stuff. Nice summary list. I bookmarked it. Thank you.

 

[Susquatch]

Here is an interesting AI response to the question "Should people be concerned about the implications of AI replacing us?"

The possibility of AI "taking over the world" is a topic of active debate among experts, policymakers, and the public. Here’s a nuanced summary of the current state:

• Leading AI researchers such as Geoffrey Hinton and others have publicly warned about the risk of advanced AI systems acting in ways humans cannot control, potentially leading to catastrophic scenarios. They stress the need for global cooperation and safeguards to prevent such outcomes.
• Some of these experts have gone so far as to compare the risks of advanced AI to other transformative technologies like nuclear weapons, emphasizing the potential for unintended consequences if development outpaces safety measures.

Short and medium term risks unclude:

Job displacement due to automation, with entire industries (such as manufacturing and administrative support) facing significant disruption.

Algorithmic bias, privacy violations, and the risk of AI-driven surveillance or misinformation campaigns.

Cybersecurity threats: AI systems can be manipulated by adversaries, creating vulnerabilities across training, deployment, and operation stages. Reports by NIST highlight these as immediate and growing risks.

Socioeconomic inequality: Concentration of AI benefits and wealth may exacerbate existing divides within and between countries.

​

While some experts argue that concerns about AI leading to human extinction are speculative, others believe the risks—however small—are significant enough to require urgent action. Surveys indicate that AI experts, on average, estimate a low but real probability (around 3% by 2100) for catastrophic "takeover"-like scenarios.​

Others, like Yann LeCun, caution that fears about superintelligent AI are premature, noting that current AI systems are nowhere near the capability to pose existential threats.​

​

Public opinion is mixed: in recent surveys, 43% of U.S. adults believe AI is more likely to harm them than benefit them, while a majority of AI experts feel the technology will be personally beneficial.​

Many in the public emphasize uncertainty, with a significant percentage unsure about AI's broader impact.​

​

Today's AI systems are powerful in narrow domains but lack general intelligence or autonomy. Most experts agree that, at present, AI serves mainly as a tool to augment human activity rather than replace humans entirely.​

The real-world scenario of conscious AI "taking over" remains in the realm of science fiction for now, but experts urge continued vigilance, research, and policy action to anticipate rapid advances and mitigate risks.​

Humans should not panic about AI taking over the world imminently, but there are legitimate reasons to be concerned about how AI is developed and deployed. Vigilant risk assessment, international cooperation, regulatory oversight, and the prioritization of AI safety research are considered essential by most experts to ensure future AI advances benefit humanity without existential downsides.

 

[Elektrishun]

Humans should not panic about AI taking over the world imminently, but there are legitimate reasons to be concerned about how AI is developed and deployed. Vigilant risk assessment, international cooperation, regulatory oversight, and the prioritization of AI safety research are considered essential by most experts to ensure future AI advances benefit humanity without existential downsides.

And there's where it all falls apart. .

If anyone is interested in listening to a long form discussion of the potential dangers of AI I would suggest the following:

Roman Yampolskiy: Dangers of Superintelligent AI | Lex Fridman Podcast #431​

One of the topics discussed is an AI tailor made for the individual. Like a multi-verse. Avoids the conflict of humans trying to agree on what AI should and shouldn't be. Your AI reflects you with all your own beliefs, ethics, etc.

People who live for online gaming and similar digital worlds are embracing this technology on a different level. Not just a tool. Who needs humans when you got your trusted digital friend.

 

[Susquatch]

I have no fondness for Government oversight, but would be pleased to support any motion that required AI generated content to carry a label saying just that. So long as it does not get confused with real world, real issues, or real (so-called) News.

Would you really trust such a law? The only people who would respect such a law are the honest ones you don't need to worry about anyway.

The crooks, shysters, cheaters, liers, and con-men wouldn't care less about faking whatever they want.

 

[Gennady]

you got your trusted digital friend

It is not your friend. It is you!!! good AI chat partner is able to adjust itself to your style, vocabulary, way of thinking , so - it is just your "alter ego". This is not a friendship.

 

[Elektrishun]

It is not your friend. It is you!!! good AI chat partner is able to adjust itself to your style, vocabulary, way of thinking , so - it is just your "alter ego". This is not a friendship.

I know that
You know that
But if "they" wanted to create the perfect seductress, the master manipulator, that's how to do it. Maybe that's the intentional/unintentional end - we all become manipulated by our digital selves into becoming docile slugs. Let Jesus AI take the wheel.

Take a moment and read through some Reddit discussions about AI. One individual described Claude as "life changing". Now, is it a real person or someone "selling" Claude?

Edit - Of course, we could be there already and don't know it.

 

[Tom O]

I think that the company’s replacing humans with AI or robotic assembly line should be taxed appropriately to help with the support of the workers let go.

 

[Gennady]

One individual described Claude as "life changing"

Paid version really is. For coding . I tried that. I asked to create quite complex and functionally rich program for Arduino, was asking for additional features, refactoring and plenty of other stuff. I just loaded that into Arduino and it worked (small twicking included, LOL ). Afte that I asked to port it to PIC18 and Claude did it. So Instead of coding and coding and debugging and coding again I was concentrating on loading AI with features. That is a huge step forward. Having that "guy" as an sole-mate.... Nah

 

[Elektrishun]

Paid version really is. For coding . I tried that. I asked to create quite complex and functionally rich program for Arduino, was asking for additional features, refactoring and plenty of other stuff. I just loaded that into Arduino and it worked (small twicking included, LOL ). Afte that I asked to port it to PIC18 and Claude did it. So Instead of coding and coding and debugging and coding again I was concentrating on loading AI with features. That is a huge step forward. Having that "guy" as an sole-mate.... Nah

See, it's working already. So here I go, into The Matrix...

I think that is super cool, too. I tried to follow setting up 2 AVR128DA48 Curiosity Nanos for a project. Couldn't get it to work. Spent hours going over it and found something in the project that needed clarification.

I submitted an "issue" on GitHub on Feb. 29, 2024. Got a response Jan. 2025.

I will definitely look into Claude.

Thanks