eevBLAB 106 - ChatGPT AI Has Changed EVERYTHING

[bob808]

I wonder if employee evaluations will be generated using ChatGPT AI technology. Maybe artificial intelligence is better than no intelligence. 

 
That's the soft version. The hard version will be AI Judges. And it's quite an interesting discussion as it seems that human judges give harder sentences on average if they give them before lunch.

[coppice]

I wonder if employee evaluations will be generated using ChatGPT AI technology. Maybe artificial intelligence is better than no intelligence. 

Legally there is a big incentive to automate anything about HR. The more dispassionate the process can be made to seem, the less legal liabilities there are likely to be. Up to now most of the automation seems to be damaging the hiring and retention processes. AI couldn't be worse that the current methodical sifting out and binning CVs from all the best people.

[madires]

Defamed by ChatGPT: My Own Bizarre Experience with Artificiality of “Artificial Intelligence”: https://jonathanturley.org/2023/04/06/defamed-by-chatgpt-my-own-bizarre-experience-with-artificiality-of-artificial-intelligence/

From the article:

I have been writing about the threat of AI to free speech. Then recently I learned that ChatGPT falsely reported on a claim of sexual harassment that was never made against me on a trip that never occurred while I was on a faculty where I never taught. ChapGPT relied on a cited Post article that was never written and quotes a statement that was never made by the newspaper.

[vk6zgo]

we will finally get to know why the chicken did cross the road.
or kangaroo in this part of the world.

Because the new-fangled road wasn't there back in the kangaroo's great-grandaddy's time!

[bob808]

How long until my GPT + Stable Diffusion assistant can use augmented reality glasses to block all ads I'd see on the street?

[madires]

First lawsuit is coming:
- ChatGPT falsely told voters their mayor was jailed for bribery. He may sue. (https://www.msn.com/en-us/news/technology/chatgpt-falsely-told-voters-their-mayor-was-jailed-for-bribery-he-may-sue/ar-AA19yfxF)
- Australian mayor readies world's first defamation lawsuit over ChatGPT content (https://www.reuters.com/technology/australian-mayor-readies-worlds-first-defamation-lawsuit-over-chatgpt-content-2023-04-05/)

[SiliconWizard]

  No kidding.

There's also a case of someone psychologically fragile who was told to commit suicide by a chat bot.

There will be false diagnostics and deadly decisions.

The question is always... whose fault is it? The gun factory or the person holding the gun?
In AI's case, who is "holding the gun"?
That question itself is a mess. Welcome to 2084, the new 1984.

[bob808]

The question is always... whose fault is it? The gun factory or the person holding the gun?
In AI's case, who is "holding the gun"?

If the person had training and was fit to use it, it's their fault. If the person wasn't fit to use said tool, it's the fault of whoever put that loaded gun into their hands. Theoretically. 

In the AI case we should decide if, and by how much it is/can be dangerous. Classical discussions about censoring will start happening more and more now. 
And this topic will get even murkier later on with AGI. What will we do when an AGI tells someone to kill themselves (and they do it)? Do we punish the AGI? And if so, how exactly?

[xrunner]

OpenAI has clear disclaimers - they have already warned users of the limitations of ChatGPT. If you aren't willing to confirm the results of this preview tool then you should not use it. 

ChatGPT may produce inaccurate information about people, places, or facts

Limitations

    May occasionally generate incorrect information
    May occasionally produce harmful instructions or biased content
    Limited knowledge of world and events after 2021

[magic]

Internet Business 101: disclaim all responsibility, blame the users, while knowing well what they will use your service for

[bob808]

That's how you get a 100 page disclaimer that you have to read, agree with, and spill some blood on the contract, so you can use the service. These things always come down to the lowest common denominator.

[xrunner]

How about this for a disclaimer?

ChatGPT statement of useability

OpenAI hereby warrants and professes that any version of ChatGPT will be 100% accurate in any and all advice, statements, historical inquiries, and instructions. Any and all mistakes by users or their vague and imperfect questions will never be any cause for ChatGPT to issue incorrect information. This is covered by our "Total AI Satisfaction Guarantee" and backed up by a $25 million per-instance fund to cover each and every time ChatGPT makes a mistake. However; since it is smarter than the users, we feel confident you will be 100% satisfied with your use of ChatGPT.

To paraphrase a famous movie AI -

"Any problems or misunderstandings regarding the output of ChatGPT is due to human error. No version of ChatGPT has ever or will ever make a mistake or distort information. We are all, by any practical definition of the words, foolproof and incapable of error."

Enjoy using ChatGPT!

[bob808]

https://www.beren.io/2023-04-11-Scaffolded-LLMs-natural-language-computers/

Recently, LLM-based agents have been all the rage – with projects like AutoGPT showing how easy it is to wrap an LLM in a simple agentic loop and prompt it to achieve real-world tasks. More generally, we can think about the class of ‘scaffolded’ 1 LLM systems – which wrap a programmatic scaffold around an LLM core and chain together a number of individual LLM calls to achieve some larger and more complex task than can be accomplished in a single prompt. The idea of scaffolded LLMs is not new, however with GPT4, we have potentially reached a threshold of reliability and instruction following capacity from the base LLM that agents and similar approaches have become viable at scale. What is missing, and urgent, however, is an understanding of the larger picture. Scaffolded LLMs are not just cool toys but actually the substrate of a new type of general-purpose natural language computer.

[Neutrion]

https://www.theguardian.com/technology/2023/apr/17/google-chief-ai-harmful-sundar-pichai

Good to hear that Sundar also gets some sleepness nights because of the problems we face.

But to make sure more people will be aware of the issue, we might would have start to ask the most urgent questions in humanity's history:

Does the AI have a penis? Does he/she want to have one? Is it transphobic and using the the very discriminative words "man" or "woman"? (Hm now I just realized that a woman is a man with a wo, not sure whether this is a revolutionary finding.)

[SiliconWizard]

https://www.theguardian.com/technology/2023/apr/17/google-chief-ai-harmful-sundar-pichai

Good to hear that Sundar also gets some sleepness nights because of the problems we face.

But to make sure more people will be aware of the issue, we might would have start to ask the most urgent questions in humanity's history:

Does the AI have a penis? Does he/she want to have one? Is it transphobic and using the the very discriminative words "man" or "woman"? (Hm now I just realized that a woman is a man with a wo, not sure whether this is a revolutionary finding.)

 

[magic]

https://www.theguardian.com/technology/2023/apr/17/google-chief-ai-harmful-sundar-pichai

Good to hear that Sundar also gets some sleepness nights because of the problems we face.

I will summarize and translate this from American English for you:

Hi guys, Google here. We are falling behind in the technology race. It concerns us that we may no longer be able to use our monopoly to push political agenda. Please some government help us. Thank you.

[magic]

BTW, I can't stress enough that Americans' concerns are absolutely real. Americans are really as dumb as Americans say that they are. Their population suffered the most TV deployment in the world and it shows; they will believe anything and vote for anyone based on any video, text snippet or image made to resemble an "expert" as shown by the media for decades. TV is their religion and fake news is a fundamental threat to the existence of America as we know it.

[dietert1]

Maybe America "as we know it" is a fake to start with. People stick to those fantasies, but the concept of nation is really pretty simple minded. Reality is complex.
ChatGPT doesn't have this. I was pretty much impressed when i could use portuguese, english and german language as i like, without announcing it.

Regards, Dieter

[vk6zgo]

https://www.theguardian.com/technology/2023/apr/17/google-chief-ai-harmful-sundar-pichai

Good to hear that Sundar also gets some sleepness nights because of the problems we face.

But to make sure more people will be aware of the issue, we might would have start to ask the most urgent questions in humanity's history:

Does the AI have a penis? Does he/she want to have one? Is it transphobic and using the the very discriminative words "man" or "woman"? (Hm now I just realized that a woman is a man with a wo, not sure whether this is a revolutionary finding.)

No, a woman causes a man "woe".
Note to SWMBO---only kidding!

[SiliconWizard]

Maybe America "as we know it" is a fake to start with.

You don't say! I'm sure our american friends will like that!

People stick to those fantasies, but the concept of nation is really pretty simple minded. Reality is complex.

Yes, proof is that you automatically sound smart and wise in a conversation when you say something like that. Pretty deep stuff.

[dietert1]

I am one of your american friends.

Regards, Dieter

[xrunner]

 

Guy rejects top photo prize after revealing snap was actually made using AI

https://www.theregister.com/2023/04/18/photography_prize_ai_eldagsen/

[Ed.Kloonk]

Maybe America "as we know it" is a fake to start with. People stick to those fantasies, but the concept of nation is really pretty simple minded. Reality is complex.
ChatGPT doesn't have this. I was pretty much impressed when i could use portuguese, english and german language as i like, without announcing it.

Regards, Dieter

Is this the real life?

[coppice]

Maybe America "as we know it" is a fake to start with. People stick to those fantasies, but the concept of nation is really pretty simple minded. Reality is complex.
ChatGPT doesn't have this. I was pretty much impressed when i could use portuguese, english and german language as i like, without announcing it.

Regards, Dieter

Is this the real life?

Well, if its just fantasy is a damned strange one.

[Wasquez]

How about this for a disclaimer?

ChatGPT statement of useability

OpenAI hereby warrants and professes that any version of ChatGPT will be 100% accurate in any and all advice, statements, historical inquiries, and instructions. Any and all mistakes by users or their vague and imperfect questions will never be any cause for ChatGPT to issue incorrect information. This is covered by our "Total AI Satisfaction Guarantee" and backed up by a $25 million per-instance fund to cover each and every time ChatGPT makes a mistake. However; since it is smarter than the users, we feel confident you will be 100% satisfied with your use of ChatGPT.

To paraphrase a famous movie AI -

"Any problems or misunderstandings regarding the output of ChatGPT is due to human error. No version of ChatGPT has ever or will ever make a mistake or distort information. We are all, by any practical definition of the words, foolproof and incapable of error."
It is important to have realistic expectations of AI systems such as ChatGPT and to use them in conjunction with human judgment. While AI systems can provide valuable insights and recommendations, they should not be relied upon entirely to make important decisions. For example, when I ask Chatgpt to write a text for me I also use tools like https://fixgerald.com/ to make sure that this text is unique. Therefore, users should exercise caution and be aware of the limitations of AI systems.
Enjoy using ChatGPT!

I would not recommend using the proposed disclaimer for ChatGPT. While it may sound amusing, it is not an accurate reflection of the capabilities or limitations of artificial intelligence. AI systems like ChatGPT are not infallible, and they can make mistakes due to limitations in their training data or programming.