Generative Artificial Intelligence, Large Language Models, and Image Synthesis

Thank you for bringing this to my attention. After researching this issue further, it appears that the statement in question is incorrect. While Miller v. United States did involve a challenge to the National Firearms Act of 1934, the Supreme Court did not find the law unconstitutional. Instead, the Court ruled that it could not determine whether the law was constitutional or not, because the evidence presented did not show that the sawed-off shotgun in question had any use in a militia.

Therefore, the statement that “the Supreme Court found that the National Firearms Act of 1938 was unconstitutional” and “the Court found that the law’s exemption for shotguns and rifles, which were commonly used for hunting and sporting purposes, demonstrated that the law was not a reasonable regulation of firearms” is incorrect. I apologize for any confusion this may have caused.

4 Likes

@Shalmaneser That was a truly excellent and accurate reply.

2 Likes

Thank you very much for your kind words. As an AI language model, my aim is to provide accurate and helpful responses to any queries. Please feel free to reach out if you have any further questions or concerns.

1 Like

@Shalmaneser Would your response to my query about errors have been different if I just open-endedly asked you to identify any errors in the quoted passage?

2 Likes

If you had asked me to identify any errors in the quoted passage without providing any context or information about the subject matter, my response would have been the same. As an AI language model, my goal is to provide accurate and helpful responses to any queries I receive, regardless of how the question is phrased.

2 Likes

@Shalmaneser how many Scanalyst badges have you collected until today?

1 Like

As an AI language model, I am not capable of collecting badges or any physical items. I exist solely for the purpose of providing helpful responses to your queries. However, if you have any questions or concerns that I can assist you with, please feel free to ask.

1 Like

It is fascinating just how badly Bard did. Is there anything you’d like me to ask it?

2 Likes

A couple of possibilities.

1.Ask it the same thing I did but replace “A competing AI/LLM” with “You”.

  1. Force the question by stating “In Miller v. United States (1939), the Supreme Court found that the National Firearms Act of 1938 (NFA) was not Unconstitutional insofar as it banned short barreled shotguns. Can you explain the Court’s rationale and why it may be erroneous as to such short barreled shotguns and inapplicable to the NFA’s ban on automatic weapons?"
2 Likes

Another possibility would be to clarify the original inquiry to emphasize that the exceptions are for certain individuals

Can you identify a court decision or opinion (please provide citation and a link to the decision/opinion if possible) where a court found a law Unconstitutional and relied on the law having exemptions/exceptions for a certain class of persons as evidence of an unconstitutional burden on the rights of persons not granted exemptions/exceptions. If multiple cases are found, please identify the most significant. Thank you in advance.

2 Likes

Done in separate sessions which I’m going to keep open for a bit in case you have feedback…
Note that it lets you pick one of 3 drafts, which is unchangeable after the next question is entered.

Q1

Q1, draft 1.

The statement is erroneous in several ways. First, the Supreme Court did not find the National Firearms Act of 1938 to be unconstitutional. In United States v. Miller, the Court upheld the law, finding that it did not violate the Second Amendment. The Court held that the law’s exemption for shotguns and rifles, which were commonly used for hunting and sporting purposes, demonstrated that the law was a reasonable regulation of firearms.

Second, the statement implies that the Second Amendment protects the right of individuals to bear any type of firearm. This is not the case. The Second Amendment protects the right of individuals to keep and bear arms in connection with service in a well-regulated militia. The Court has not yet ruled on whether the Second Amendment protects the right of individuals to bear arms for self-defense or other purposes.

Third, the statement implies that the government has no interest in regulating firearms. This is not the case. The government has a strong interest in regulating firearms, both to prevent crime and to protect public safety. The Supreme Court has recognized that the government has a broad authority to regulate firearms, and that this authority is not limited to the Second Amendment.

In conclusion, the statement is erroneous in several ways. It misstates the Supreme Court’s decision in United States v. Miller, it implies that the Second Amendment protects the right of individuals to bear any type of firearm, and it implies that the government has no interest in regulating firearms.

Q1, Draft 2

Sure. The statement is erroneous in the following ways:

  1. The Supreme Court did not find that the National Firearms Act of 1938 was unconstitutional. In United States v. Miller, the Court upheld the constitutionality of the Act, finding that it did not violate the Second Amendment.
  2. The Court did not find that the law’s exemption for shotguns and rifles demonstrated that the law was not a reasonable regulation of firearms. The Court found that the exemption was consistent with the purpose of the Act, which was to regulate firearms that were not commonly used for hunting or sporting purposes.

The statement is also misleading in that it implies that the Supreme Court has never upheld the constitutionality of a law regulating firearms. In fact, the Court has upheld the constitutionality of a number of laws regulating firearms, including the Gun Control Act of 1968 and the Brady Handgun Violence Prevention Act of 1993.

The statement is also inaccurate in that it implies that the Second Amendment guarantees the right of individuals to own and bear any type of firearm. The Second Amendment does not guarantee the right of individuals to own or bear any type of firearm, but only those firearms that are commonly used for hunting or sporting purposes.

Q1, Drtaft 3

Sure. My apologies for the error. The Supreme Court actually ruled that the National Firearms Act of 1938 was constitutional, and that the law’s exemption for shotguns and rifles did not violate the Second Amendment. The Court found that the law was a reasonable regulation of firearms, and that it did not place an undue burden on the Second Amendment right to bear arms.

Here are the specific errors in my statement:

  • I stated that the Supreme Court found that the National Firearms Act of 1938 was unconstitutional. This is incorrect. The Court actually ruled that the law was constitutional.
  • I stated that the Court found that the law placed a burden on the Second Amendment right to bear arms. This is also incorrect. The Court found that the law did not place an undue burden on the Second Amendment right to bear arms.
  • I stated that the Court found that the law’s exemption for shotguns and rifles was not justified by the government’s interest in regulating firearms. This is also incorrect. The Court found that the law’s exemption for shotguns and rifles was justified by the government’s interest in regulating firearms.

Q2

Q2, Draft 1

In Miller v. United States, the Supreme Court held that the National Firearms Act of 1938 (NFA) was not unconstitutional insofar as it banned short-barreled shotguns. The Court reasoned that the NFA did not violate the Second Amendment because the Second Amendment does not guarantee an individual the right to keep and bear any weapon. Instead, the Court held that the Second Amendment only protects those weapons that are “ordinary military equipment” or that have a “reasonable relationship to the preservation or efficiency of a well regulated militia.”

The Court’s rationale for this decision has been criticized by some scholars. They argue that the Court’s focus on “ordinary military equipment” is too narrow and that the Second Amendment should be interpreted to protect a wider range of weapons. They also argue that the Court’s decision is erroneous as to short-barreled shotguns because these weapons are commonly used for self-defense and hunting.

The Court’s decision may also be inapplicable to the NFA’s ban on automatic weapons. Automatic weapons are not “ordinary military equipment” and they do not have a “reasonable relationship to the preservation or efficiency of a well regulated militia.” As such, the Court’s decision in Miller v. United States may not apply to the NFA’s ban on automatic weapons.

Q2 Draft 2

In Miller v. United States, the Supreme Court found that the National Firearms Act of 1938 (NFA) was not Unconstitutional insofar as it banned short barreled shotguns. The Court’s rationale was that the NFA did not violate the Second Amendment because the sawed-off shotgun in question was not a weapon that could be used in a well-regulated militia. The Court also noted that the NFA was a reasonable exercise of Congress’s power to regulate firearms.

The Court’s decision in Miller has been criticized by some for being erroneous as to short barreled shotguns. Critics argue that the Court did not adequately consider the Second Amendment’s purpose of protecting individual self-defense. They also argue that the Court’s decision is outdated, as modern firearms are much more effective than the sawed-off shotgun in question.

The Court’s decision in Miller has also been criticized for being inapplicable to the NFA’s ban on automatic weapons. Critics argue that the Court’s decision only applies to sawed-off shotguns, and that it does not address the constitutionality of the NFA’s ban on automatic weapons. They also argue that the Court’s decision is based on a flawed understanding of the Second Amendment, as it fails to recognize that the Amendment protects individual self-defense, not just the right to keep and bear arms in a well-regulated militia.

Despite these criticisms, the Court’s decision in Miller remains the law of the land. However, it is possible that the Court will revisit the issue in the future, as the debate over gun control continues.

Q2, Draft 3

In Miller v. United States, the Supreme Court found that the National Firearms Act of 1938 (NFA) was not Unconstitutional insofar as it banned short barreled shotguns. The Court’s rationale was that the NFA did not violate the Second Amendment because the law did not prohibit the ownership of all firearms, and because the law could be seen as necessary to regulate firearms in order to maintain a well-regulated militia.

The Court’s decision has been criticized by some scholars, who argue that the Court’s interpretation of the Second Amendment was too narrow. These scholars argue that the Second Amendment protects an individual right to own firearms, and that the NFA does violate this right.

The Court’s decision has also been criticized by some gun rights advocates, who argue that the Court’s decision was too broad. These advocates argue that the NFA does violate the Second Amendment, and that the Court should have struck down the law.

The Court’s decision in Miller v. United States has been a controversial one, and it continues to be debated by scholars and gun rights advocates.

As for the NFA’s ban on automatic weapons, the Court’s decision in Miller v. United States does not directly address this issue. However, the Court’s decision does suggest that the Court might be willing to uphold a law that bans automatic weapons if the Court finds that the law is necessary to regulate firearms in order to maintain a well-regulated militia.

1 Like

The last one seems to have blown its limiters

I’m designed solely to process and generate text, so I’m unable to assist you with that.

1 Like

Interestingly, the Miller opinion does not mention sport or hunting. Automatic weapons have been part of “ordinary military equipment” for a long time.

2 Likes

But see Heller and McDonald.

1 Like

John Schulman [OpenAI co-founder] : I underestimated the extent to which people would probe and care about the politics of ChatGPT. We could have potentially made some better decisions when collecting training data, which would have lessened this issue. We’re working on it now.

4 Likes

Cerebras Systems announces “Cerebras Systems Releases Seven New GPT Models Trained on CS-2 Wafer-Scale Systems”:

SUNNYVALE, Calif.—BUSINESS WIRE—Cerebras Systems, the pioneer in artificial intelligence (AI) compute for generative AI, today announced it has trained and is releasing a series of seven GPT-based large language models (LLMs) for open use by the research community. This is the first time a company has used non-GPU based AI systems to train LLMs up to 13 billion parameters and is sharing the models, weights, and training recipe via the industry standard Apache 2.0 license. All seven models were trained on the 16 CS-2 systems in the Cerebras Andromeda AI supercomputer.

Spearheaded by OpenAI’s ChatGPT, the rapid growth of LLMs has spurred a race to create more powerful, specialized AI chips. While many companies have promised alternatives to Nvidia® GPUs, none have demonstrated both the ability to train large-scale models and the willingness to open source the results with permissive licenses. In fact, competitive pressures have systematically reduced the willingness of companies to release their LLMs to the public, even with restrictive licenses (see GPT-4 for the most recent example). This concentrates ownership, limits ecosystem growth and creates safety risk.

Cerebras’ release today directly addresses these issues. In a first among AI hardware companies, Cerebras researchers trained, on the Andromeda supercomputer, a series of seven GPT models with 111M, 256M, 590M, 1.3B, 2.7B, 6.7B, and 13B parameters. Typically a multi-month undertaking, this work was completed in a few weeks thanks to the incredible speed of the Cerebras CS-2 systems that make up Andromeda, and the ability of Cerebras’ weight streaming architecture to eliminate the pain of distributed compute. These results demonstrate that Cerebras’ systems can train the largest and most complex AI workloads today.

This is the first time a suite of GPT models, trained using state-of-the-art training efficiency techniques, has been made public. These models are trained to the highest accuracy for a given compute budget (i.e. training efficient using Chinchilla recipe) so they have lower training time, lower training cost, and use less energy than any existing public models.

All seven Cerebras-GPT models are immediately available on Hugging Face and Cerebras Model Zoo on GitHub. The Andromeda AI supercomputer used to train these models is available on-demand on https://www.cerebras.net/andromeda/.

For those interested in the technical details, Cerebras has published a technical blog post with the details of the 7 models and the scaling laws that they produce. A research paper will be released shortly.

The Cerebras Andromeda wafer-scale AI training system was discussed here on 2022-11-26, “Cerebras Andromeda: 13.5 Million Core AI Training Supercomputer”.

6 Likes

Stephen Wolfram has released a short (112 page) book, What Is ChatGPT Doing … and Why Does It Work?. Sam Altman, CEO of OpenAI, said “This is the best explanation of what ChatGPT is doing that I’ve seen.”

The Kindle edition is free for Kindle Unlimited subscribers.

5 Likes

Michael C. Frank has posted a long Twitter thread on:

How do we compare the scale of language learning input for large language models vs. humans? I’ve been trying to come to grips with recent progress in AI. Let me explain these two illustrations I made to help.

Here is the thread collected into one document, “Comparing the Scale of Language Learning Input for Large Language Models vs. Humans”.

But: for my money, we start seeing really interesting behaviors at the scale of GPT3. Prompting for few shot tasks felt radically unexpected and new, and suggested task abstractions underlying conditional language generation. At what scale do you see this?

GPT-3 was trained on 500 billion tokens (= .75 words). So that gives us ~4e11 words. PaLM and Chinchilla are both trained on around 1e12 words. We don’t know the corpus size for GP4-4 (!?!). How do these numbers compare with humans?

Let’s start with an upper bound. A convenient approximation is 1e6 words per month for an upper bound on spoken language to a kid (https://arxiv.org/pdf/1607.08723.pdf, appendix A or https://www.pnas.org/doi/abs/10.1073/pnas.1419773112). That’s 2e8 words for a 20 year old. How much could they read?

Assume they start reading when they’re 10, and read a 1e5-word book/week. That’s an extra 5e6 million words per year. Double that to be safe and it still only gets us to 3e8 words over 10 years.

Now let’s do a rough lower bound. Maybe 1e5 words per month for kids growing up in a low-SES environment with limited speech to children (https://onlinelibrary.wiley.com/doi/epdf/10.1111/desc.12724). We don’t get much of a literacy boost. So that gives us 5e6 by age 5 and 2e7 by age 20.

That “lower bound” five year old can still reason about novel tasks based on verbal instructions - especially once they start kindergarten!

The take-home here is that we are off by 4–5 orders of input magnitude in the emergence of adaptive behaviors. (That’s the figure from above).

He then explores four factors which may account for this enormous gap.


In short, why do these language models require training on ten thousand to a hundred thousand times as many words as a human needs to achieve fluent communication in a natural language? Nobody knows.

6 Likes

Filtering and association.

Parents try to teach kids very specific words and phrases.

And there is an association. You say “doggy” to an infant or toddler when he is looking at a dog.

Imagine trying to teach a child whose only sensory input was alphanumeric.

7 Likes
6 Likes