People often think that the Turing Test was meant to be a test of intelligence, but it wasn't. It was an attempt at creating a definition of "intelligent behavior".

The operative word being "attempt"

 

The operative word being "attempt"

I'd have to go back and read his original paper (it's been many years since I did that and I wasn't thinking in these terms at the time, my focus was completely different), but I'm curious to see to what degree he seemed to be aware that he was making a distinction between "intelligence" and "intelligent behavior" (though I recall him using the latter phrase pretty consistently), but more importantly, whether he made it clear that he was coming up with a definition as a consequence of there really not being one, or whether he thought his definition was merely in concert with common understanding.

 

This is interesting
I think
https://www.theguardian.com/technol...bot-turned-against-artificial-intelligence-ai

 

Seems others are asking how to test and quantify is ai intelegent
https://www.nature.com/articles/d41586-023-02361-7

 

Or may be it is ...

https://www.theregister.com/2023/07/31/llms_paper_nature/

 

https://www.fastcompany.com/90929549/google-jigsaw-toxic-speech-ai
Google’s Jigsaw was trying to fight toxic speech with AI. Then the AI started talking
OpenAI, Anthropic, and others are using Jigsaw’s Perspective—designed to moderate toxic human speech—to evaluate their large language models. What could go wrong?

For her part, Vasserman is realistic about the limits of AI when it comes to detecting harmful speech, whoever—or whatever—is talking. AI can classify and produce words, but for now it doesn’t know the world in which those words operate and can’t know the damage toxic language can do.

“I think we are slowly but surely generally coming to a consensus around, these are the different types of problems that you want to be thinking about, and here are some techniques,” she says. “But I think we’re still—and we’ll always be—far from having it fully solved.”

 

https://www.dailystar.co.uk/news/weird-news/ai-replacing-psychics-fortune-telling-30659524

According to website Medium: "In the digital age the convergence of ancient wisdom and cutting-edge technology has given rise to new possibilities.

"One such innovation is AI tarot reading where the age-old practice of tarot cards meets the power of artificial intelligence.


"AI tarot reading involves using sophisticated algorithms and machine learning techniques to interpret and generate tarot readings.

"Through natural language processing and data analysis AI algorithms aim to mimic the intuition and insight traditionally associated with human tarot readers.

 

LLMs like GPT and Bard can be manipulated and hypnotized
Hypnotized LLMs can help leak confidential financial information, generate malicious code and even cross red lights.

​

 

https://www.theguardian.com/world/2023/aug/10/pak-n-save-savey-meal-bot-ai-app-malfunction-recipes

One recipe it dubbed “aromatic water mix” would create chlorine gas. The bot recommends the recipe as “the perfect nonalcoholic beverage to quench your thirst and refresh your senses”.

“Serve chilled and enjoy the refreshing fragrance,” it says, but does not note that inhaling chlorine gas can cause lung damage or death.

New Zealand political commentator Liam Hehir posted the “recipe” to Twitter, prompting other New Zealanders to experiment and share their results to social media. Recommendations included a bleach “fresh breath” mocktail, ant-poison and glue sandwiches, “bleach-infused rice surprise” and “methanol bliss” – a kind of turpentine-flavoured french toast.

 

https://arxiv.org/abs/2308.02312
Who Answers It Better? An In-Depth Analysis of ChatGPT and Stack Overflow Answers to Software Engineering Questions

Over the last decade, Q&A platforms have played a crucial role in how programmers seek help online. The emergence of ChatGPT, however, is causing a shift in this pattern. Despite ChatGPT's popularity, there hasn't been a thorough investigation into the quality and usability of its responses to software engineering queries. To address this gap, we undertook a comprehensive analysis of ChatGPT's replies to 517 questions from Stack Overflow (SO). We assessed the correctness, consistency, comprehensiveness, and conciseness of these responses. Additionally, we conducted an extensive linguistic analysis and a user study to gain insights into the linguistic and human aspects of ChatGPT's answers. Our examination revealed that 52% of ChatGPT's answers contain inaccuracies and 77% are verbose. Nevertheless, users still prefer ChatGPT's responses 39.34% of the time due to their comprehensiveness and articulate language style. These findings underscore the need for meticulous error correction in ChatGPT while also raising awareness among users about the potential risks associated with seemingly accurate answers.

Users get tricked by appearance.

Our user study results show
that users prefer ChatGPT answers 34.82% of the time. However,
77.27% of these preferences are incorrect answers. We believe this
observation is worth investigating. During our study, we observed
that only when the error in the ChatGPT answer is obvious, users
can identify the error. However, when the error is not readily verifiable or requires external IDE or documentation, users often fail
to identify the incorrectness or underestimate the degree of error
in the answer. Surprisingly, even when the answer has an obvious error, 2 out of 12 participants still marked them as correct
and preferred that answer. From semi-structured interviews, it is
apparent that polite language, articulated and text-book style answers, comprehensiveness, and affiliation in answers make completely wrong answers seem correct. We argue that these seemingly correct-looking answers are the most fatal. They can easily
trick users into thinking that they are correct, especially when they
lack the expertise or means to readily verify the correctness. It is
even more dangerous when a human is not involved in the generation process and generated results are automatically used elsewhere by another AI. The chain of errors will propagate and have
devastating effects in these situations. With the large percentage
of incorrect answers ChatGPT generates, this situation is alarming.
Hence it is crucial to communicate the level of correctness to users

 

Just one more example of the trend toward people wanting to let the tool do the thinking for them instead of putting forth the effort to gain a solid grounding in the material so that the tool is simply augmenting their abilities.

 

Probably the best definition of the technology that I've heard so far:

Top physicist says chatbots are just ‘glorified tape recorders,’ and predicts a different computing revolution is ahead

​

“I don’t think A.I. is going to take over. There’s no independent entity that wants to conquer you,” David Ferrucci, A.I. scientist and the creator of IBM Watson, told Observer recently. For now, A.I. technology is not nearly as skilled as humans. For example, chatbots cannot discern true from false, Kaku said. “That has to be put in by a human.” (Companies like Anthropic are in the process of developing language models that are aligned with human values.)

 

Probably the best definition of the technology that I've heard so far:

Top physicist says chatbots are just ‘glorified tape recorders,’ and predicts a different computing revolution is ahead
View attachment 300587​

Even Michio Kaku (blinded by the bright lights of fame) is occasionally right about technology.

Actual experts of quantum computing all say don't listen to him about quantum computing, AI applications and a possible AI future.
https://scottaaronson.blog/?p=7321

In the spirit, perhaps, of the TikTokkers who eat live cockroaches or whatever to satisfy their viewers, I decided to oblige loyal Shtetl-Optimized fans by buying Quantum Supremacy and reading it. So I can now state with confidence: beating out a crowded field, this is the worst book about quantum computing, for some definition of the word “about,” that I’ve ever encountered.

Admittedly, it’s not obvious why I’m reviewing the book here at all. Among people who’ve heard of this blog, I expect that approximately zero would be tempted to buy Kaku’s book, at least if they flipped through a few random pages and saw the … level of care that went into them. Conversely, the book’s target readers have probably never visited a blog like this one and never will. So what’s the use of this post?

Kaku’s slapdash “book,” and the publicity campaign around it, represents a noxious step backwards. The wonder of it, to me, is Kaku holds a PhD in theoretical physics. And yet the average English major who’s written a “what’s the deal with quantum computing?” article for some obscure link aggregator site has done a more careful and honest job than Kaku has. That’s setting the bar about a millimeter off the floor. I think the difference is, at least the English major knows that they’re supposed to call an expert or two, when writing about an enormously complicated subject of which they’re completely ignorant.

 

https://hedgehogreview.com/web-features/thr/posts/the-elon-effect

 

https://www.npr.org/2023/08/16/1194...on-against-openai-as-copyright-tensions-swirl

The New York Times and OpenAI could end up in court.

Lawyers for the newspaper are exploring whether to sue OpenAI to protect the intellectual property rights associated with its reporting, according to two people with direct knowledge of the discussions.
...
If OpenAI is found to have violated any copyrights in this process, federal law allows for the infringing articles to be destroyed at the end of the case.

In other words, if a federal judge finds that OpenAI illegally copied the Times' articles to train its AI model, the court could order the company to destroy ChatGPT's dataset, forcing the company to recreate it using only work that it is authorized to use.

 

https://www.npr.org/2023/08/16/1194...on-against-openai-as-copyright-tensions-swirl

I think this is going a bit overboard. It's like saying that anyone that has based their opinions, even in part, on something they read in the NYT has to forget what they read and form new opinions.

They put it out there for the world to see. Presumably they don't have a problem with people learning from it, so they shouldn't be able to oppose an LLM learning from it.

From a different perspective, what is it that these LLM are actually "learning" from it? Merely a bunch of statistics about how frequently words appear in close proximity to other groups of words. So if someone were to analyze publicly available NYT content and say that the word 'widget' frequently appears shortly after 'market' and 'economics' and 'free', would they contend that that violated the intellectual property rights of their reporting?

 

If there is anybody to whom I would listen on this topic...

 

https://www.physicsforums.com/insights/why-chatgpt-is-not-reliable/

start with the simple fact: ChatGPT is not a reliable answerer of questions. To try to explain why from scratch would be a heavy lift, but fortunately, Stephen Wolfram has already done the heavy lifting for us in his article, “What is ChatGPT Doing… and Why Does It Work?

https://writings.stephenwolfram.com/2023/02/what-is-chatgpt-doing-and-why-does-it-work/

It’s Just Adding One Word at a Time

 

If there is anybody to whom I would listen on this topic...

One thing that this, and so many other, explanations gloss over or ignore completely is reality. He was fairly explicit on describing how humans should interact with these language models. But the reality is that the vast majority of people are not going to interact with them the way they should -- they are going to provide poorly formed prompts and accept whatever it spews out as authoritative. It doesn't matter what we do to try to change that, that is what is going to happen.

 

One thing that this, and so many other, explanations gloss over or ignore completely is reality. He was fairly explicit on describing how humans should interact with these language models. But the reality is that the vast majority of people are not going to interact with them the way they should -- they are going to provide poorly formed prompts and accept whatever it spews out as authoritative. It doesn't matter what we do to try to change that, that is what is going to happen.

Yup. Just like the old adage goes: "garbage in, garbage out".