This is not your average Joe on the street. This is Stephen Wolfram, the mind behind mahtematic and more than one solver programs. He is behind Wolfram Alpha and is the inventor of "A New Kind of Science" https://www.wolframscience.com/nks/

 

This is not your average Joe on the street. This is Stephen Wolfram, the mind behind mahtematic and more than one solver programs. He is behind Wolfram Alpha and is the inventor of "A New Kind of Science" https://www.wolframscience.com/nks/

Stephen is still human and sometimes comes across as eccentric and a bit of a crackpot when pushing some of his novel ideas (like "New Kind of Science”) as mainstream science. That's said, he obviously see's LLM's for what they are and notes their limitations rooted in not understanding anything but how to select the next word.

 

This is not your average Joe on the street. This is Stephen Wolfram, the mind behind mahtematic and more than one solver programs. He is behind Wolfram Alpha and is the inventor of "A New Kind of Science" https://www.wolframscience.com/nks/

So? I know exactly who it is and I never said that HE uses it the wrong way. Quite the opposite, I said that HE pretty explicitly states how it SHOULD be used, namely form a query and then analyze the response and, if necessary, revise the query and repeat the process. But he, like most people, completely ignores the reality of how it WILL be used. How long did it take before a bunch of different lawyers got caught just feeding a question to an LLM and then submitting the responses to a court as part of their legal filing without ever asking if any of the cases cited even exist? Did any of these law firms do what they should have? Of course not. But that is beside the point, this is what some fraction of law firms are going to do.

This is a problem that manifests itself over and over and over in just about every field imaginable. People that develop or champion a technology either ignore how people will actually use it improperly, whether intentionally or not, or they assume that all that is required is to explain the proper way that the technology should be used and that somehow that is sufficient to address the problem. In many cases, they then act surprised when they learn that people are not using it properly, when that is not only a possible result, but a virtually guaranteed certain result.

The internet protocols themselves are a shining example -- the premise behind nearly all of the protocols upon which everything else is based assume that people will play nice and follow the protocols. So virtually no thought to security was incorporated but, instead, had to be bolted on at the end and somehow made to work while riding on top of an insecure foundation.

Another example is a password. There's no shortage of people telling everyone what makes a good, strong password, including never writing it down, but then they shake their heads in dismay when they learn how many people use poor passwords that are written down in obvious places. There are banks that issue debit cards and don't give the user the choice to come up with their own PIN based on the very reasonable premise that most people will use the same PIN on multiple cards and that this is not ideal. So, instead, they "solve" that problem by issuing the user a randomly generated PIN, even though just a moment's reflection tells us that the inevitable consequence is that many users are going to write the PIN on the back of the card so that they can find it easily every time they need it since there is no way they are going to remember a dozen different PINs for a dozen different cards. When that's pointed out to them, their answer is that, "Well, they shouldn't do that." Doesn't matter what they shouldn't do, what matters is what they will do.

Yet another famous example is the breaking of the Enigma machine. By and large, the Germans were aware of its weaknesses and had protocols in place that, had they been followed, would almost certainly have prevented the Allies from achieving any meaningful breaks into the system. But it relied on young, poorly trained operators who were often freezing and hungry and under fire following the protocols; enough of them did not do so such that each day's settings were often determined within six hours of the daily key change. Was it the fault of the operators? Not really. Not fundamentally. The fault was with the designers who imposed an unrealistic belief that all users would obey whatever protocols were established, instead of spending any time learning how real operators and real situations are actually going to behave.

It's really nothing more than an example of what von Clausewitz said in On War regarding military theories and doctrine. He pointed out that there are three principal entities, the government, the military, and the people, and theories and doctrines must recognize the relationships between them. Any theory or doctrine that attempts to either ignore one of those relationships, or that attempts to impose an artificial relationship between two them, is so at odds with reality as to be fundamentally flawed from the beginning. The same is true in most realms -- if your use case relies on people doing what they should do instead of what they will do, your use case is so at odds with reality as to be largely useless.

 

https://venturebeat.com/ai/us-dod-ai-chief-on-llms-i-need-hackers-to-tell-us-how-this-stuff-breaks/

On the main stage at the DEF CON security conference in a Friday afternoon session (Aug. 11), Craig Martell, chief digital and AI officer at the U.S. Defense Department (DoD), came bearing a number of key messages.

First off, he wants people to understand that large language models (LLMs) are not sentient and aren’t actually able to reason.


Martell and the DoD also want more rigor in model development to help limit the risks of hallucination — wherein AI chatbots generate false information. Martell, who is also an adjunct professor at Northeastern University teaching machine learning (ML), treated the mainstage DEF CON session like a lecture, repeatedly asking the audience for opinions and answers.

Martell spent a lot of time during his session pointing out that LLMs don’t actually reason. In his view, the current hype cycle surrounding generative AI has led to some misplaced hype and understanding about what an LLM can and cannot do.

“We evolved to treat things that speak fluently as reasoning beings,” Martell said.
...
“I need five nines [99.999% accuracy] of correctness,” he said. “I cannot have a hallucination that says: ‘Oh yeah, put widget A connected to widget B’ — and it blows up.”

 

Calling this tech LLMs makes perfect sense. I just wish the media would refer to it also as AI as though the terms were interchangeable.

 

https://www.hollywoodreporter.com/b...i-works-not-copyrightable-studios-1235570316/

AI-Created Art Isn’t Copyrightable, Judge Says in Ruling That Could Give Hollywood Studios Pause
A federal judge on Friday upheld a finding from the U.S. Copyright Office that a piece of art created by AI is not open to protection.

 

https://www.theregister.com/2023/08/22/ibm_says_genai_can_convert/

Convert COBOL code to Java? That seems like jumping from the frying pan into the fire.

COBOL to Rust is the future, man.

 

For fun I asked ChatGPT for a way of scaling one input signal voltage range to a different range.
Check its maths. So much for artificial intelligence!

 

For fun I asked ChatGPT for a way of scaling one input signal voltage range to a different range.
Check its maths. So much for artificial intelligence!

View attachment 301094

The answer is correct within an order of magnitude.

 

For fun I asked ChatGPT for a way of scaling one input signal voltage range to a different range.
Check its maths. So much for artificial intelligence!

View attachment 301094

Not surprising. If anything, what would be surprising is that it got that close. But, if you were to just Google for this you would likely find lots of sites that had similar content, so not too surprising that stringing those words together in that order falls within the realm of highest likelihood choices of the model.

 

I'm not surprised by the answer, given ChatGPT's modus operandi, but it does reinforce the warning (that chatGPT itself gives) about not relying on its responses. I'm sure many users (and especially the media) ignore, or are unaware of, its limitations.

 

https://www.theguardian.com/books/2...d-rachel-cusks-pirated-works-used-to-train-ai

Zadie Smith, Stephen King, Rachel Cusk and Elena Ferrante are among thousands of authors whose pirated works have been used to train artificial intelligence tools, a story in The Atlantic has revealed.

More than 170,000 titles were fed into models run by companies including Meta and Bloomberg, according to an analysis of “Books3” – the dataset harnessed by the firms to build their AI tools.

Books3 was used to train Meta’s LLaMA, one of a number of large language models – the best-known of which is OpenAI’s ChatGPT – that can generate content based on patterns identified in sample texts. The dataset was also used to train Bloomberg’s BloombergGPT, EleutherAI’s GPT-J and it is “likely” it has been used in other AI models.

 

https://arxiv.org/pdf/2307.01850.pdf
Self-Consuming Generative Models Go MAD

Seismic advances in generative AI algorithms for imagery, text, and other data
types has led to the temptation to use synthetic data to train next-generation
models. Repeating this process creates an autophagous (“self-consuming”) loop
whose properties are poorly understood. We conduct a thorough analytical and
empirical analysis using state-of-the-art generative image models of three families
of autophagous loops that differ in how fixed or fresh real training data is available
through the generations of training and in whether the samples from previous
generation models have been biased to trade off data quality versus diversity. Our
primary conclusion across all scenarios is that without enough fresh real data in
each generation of an autophagous loop, future generative models are doomed to
have their quality (precision) or diversity (recall) progressively decrease. We term
this condition Model Autophagy Disorder (MAD), making analogy to mad cow
disease

 

https://www.404media.co/ai-generated-mushroom-foraging-books-amazon/
‘Life or Death:’ AI-Generated Mushroom Foraging Books Are All Over Amazon

“There are hundreds of poisonous fungi in North America and several that are deadly,” Sigrid Jakob, president of the New York Mycological Society, told me in an email. “They can look similar to popular edible species. A poor description in a book can mislead someone to eat a poisonous mushroom.”

A quick scan of Amazon’s mushroom and foraging books revealed a bunch of books likely written by ChatGPT, but are sold without any indication that they’re AI-generated and are marketed as having been written by a human when they’re very likely not.

 

https://arstechnica.com/tech-policy...-every-chatgpt-response-is-a-derivative-work/

This week, OpenAI finally responded to a pair of nearly identical class-action lawsuits from book authors—including Sarah Silverman, Paul Tremblay, Mona Awad, Chris Golden, and Richard Kadrey—who earlier this summer alleged that ChatGPT was illegally trained on pirated copies of their books.

 

https://www.404media.co/kaedim-ai-startup-2d-to-3d-used-cheap-human-labor/
Buzzy AI Startup for Generating 3D Models Used Cheap Human Labor

An artificial intelligence company, whose founder Forbes included in a 30 Under 30 list recently, promises to use machine learning to convert clients’ 2D illustrations into 3D models. In reality the company, called Kaedim, uses human artists for “quality control.” According to two sources with knowledge of the process interviewed by 404 Media, at one point, Kaedim often used human artists to make the models. One of the sources said workers at one point produced the 3D design wholecloth themselves without the help of machine learning at all.
...
What Kaedim’s artificial intelligence produced was of such low quality that at one point in time “it would just be an unrecognizable blob or something instead of a tree for example,” one source familiar with its process said. 404 Media granted multiple sources in this article anonymity to avoid retaliation.

 

Caught part of a news segment where they were interviewing someone with one of the big teachers unions about ways to use ChatGPT in the classroom. While she did at least mention issues with accuracy and the need to teach kids to verify answers, she made a statement that I thought summed things up pretty well, but not in the way she meant. After talking about how ChatGPT can lessen the drudge work and let kids focus on what really matters, she compared the benefits of using ChatGPT in the classroom to the benefits that came from introducing calculators into the classroom.

I just wanted to yell at the screen, "Yes! And THAT'S the PROBLEM!"

 

Explore the chat gpt because AI is the future.

 

Explore the chat gpt because AI is the future.

In that case, humanity is doomed.

 

In that case, humanity is doomed.

Also, I don't think ChatGTP qualifies as AI under a strict standard