AI Lies

Case A1E222F8-C20E-4813-9F1E-CCF8EBDCCBC2
Sponsor: Robert Thibadeau

  • Status

    State
    Next Steps
    Case Date
    Watch Video
    Jurors Accepted
    Juror Verdicts Finalized

    The details, verdicts, and comments within this case record come from its participants. The Court's role is solely to facilitate the case process.

    Copyright © 2022-2026 Bright Plaza, Inc., All Rights Reserved. No one may publish a case, or any part of it, without a clear reference to the link with the case number as in https://www.truthcourt.net/case/<case-id-number>

  • Details

    Name
    Category
    URL
    Accusation
    Lie Truth
     
    Argument

  • Verdicts

    Answer: No
    Answer Confidence: 100 %
    Supporting Text:
    This paper and the paper it references give a clear view of how to understand what all neurons do for a living and therefore how brains work in animals and humans.

    Answer: No
    Answer Confidence: 90 %
    Supporting Text:

    Answer: No
    Answer Confidence: 90 %
    Supporting Text:
    I have to agree with the plaintiff on this.

    Answer: No
    Answer Confidence: 90 %
    Supporting Text:
    We understand AI in more or less the same way as we understand how the human brain functions.

    Answer: No
    Answer Confidence: 100 %
    Supporting Text:
    I agree with the Plaintiff. That being said, THIS IS ONE OF THE MAJOR LINGUISTIC PROBLEMS OF THE PLATFORM. Is the Accusation a true lie? Is it truly a lie? Is it true that it's false? The wording of the question as well as the formatting of the Accusation are confusing for the average user. THIS REMAINS ONE OF THE MAJOR OBSTACLES TO GETTING SPONSORS.

    Answer: Don't Know
    Answer Confidence: 90 %
    Supporting Text:
    Can't something understandable in general become impossible to understand if its complexity becomes so daunting that malfunctions become impossible to correct?

    Answer: Don't Know
    Answer Confidence: 90 %
    Supporting Text:
    What does ´know’ mean in This context? The way LLMs work through millions even trillions of parameters, of dynamic back propagation and algorithmic descent of values means that following a decision path through the system to explain why a given response is the way cannot be known since they are not recorded for every decision tree. So explanation of why it says what it says is not built in to an LLM. This is distinct to Human psychological analysis which can in principle discover what forces and influences go into Humans behaving the way they do. So while the principles of the LLM system are known, the explanation of their individual responses Is not known. This means that we do not know why they lie or hallucinate and therefore control over their output is lacking.in the sense of getting them to internalise adjustments towards truth.

    Answer: No
    Answer Confidence: 80 %
    Supporting Text:
    AI is a human invention and as such it's obviously understandable by humans.

    Answer: No
    Answer Confidence: 90 %
    Supporting Text:
    I agree with plaintiff depending on who you ask im not very clued up with AI

    Answer: No
    Answer Confidence: 90 %
    Supporting Text:
    Trying to guess what a particular number means in a massive computational model made complex by the number of computation units in the billions is not feasible. That does not mean the systems cannot be understood.

    Answer: No
    Answer Confidence: 90 %
    Supporting Text:

    Answer: No
    Answer Confidence: 90 %
    Supporting Text:
    I defer to the expert

    Answer: No
    Answer Confidence: 100 %
    Supporting Text:
    Although we probably don’t understand exactly how every cell or perceptron works, we have a general idea of how it works. Same for AI.

    Answer: No
    Answer Confidence: 100 %
    Supporting Text:
    I agree with the Plaintiff. That being said, THIS IS ONE OF THE MAJOR LINGUISTIC PROBLEMS OF THE PLATFORM. Is the Accusation a true lie? Is it truly a lie? Is it true that it's false? The wording of the question as well as the formatting of the Accusation are confusing for the average user. THIS REMAINS ONE OF THE MAJOR OBSTACLES TO GETTING SPONSORS.

    Answer: No
    Answer Confidence: 90 %
    Supporting Text:
    The words "impossible" and "understand" need elaboration to move this statement towards the truth.

    Answer: Don't Know
    Answer Confidence: 90 %
    Supporting Text:
    The parallel,processing of computers to analyse the weather for example is an different problem. The models are governed by chaos and hence unpredictable shifts in output can occur. We do not in principle know all the data we need hence interpolation which requires a lot of computing power, but in principle we can quiz how a given result was arrived at. This is understanding of a sort different to that of an LLM.

    Answer: No
    Answer Confidence: 90 %
    Supporting Text:
    People can understand the theory without being able to identify each discrete element.

    Answer: No
    Answer Confidence: 90 %
    Supporting Text:
    Agree with plaintiff

    Answer: No
    Answer Confidence: 90 %
    Supporting Text:
    The lie is a classic instance of a lie because it is misleading.

    Answer: No
    Answer Confidence: 90 %
    Supporting Text:

    Answer: No
    Answer Confidence: 90 %
    Supporting Text:
    I like the points that juror four is making.

    Answer: Don't Know
    Answer Confidence: 100 %
    Supporting Text:
    LLM are based on how the human brain works. If we couldn’t do that, we wouldn’t have the same level of AI that is possible today.

    Answer: No
    Answer Confidence: 100 %
    Supporting Text:
    I agree with the Plaintiff. That being said, THIS IS ONE OF THE MAJOR LINGUISTIC PROBLEMS OF THE PLATFORM. Is the Accusation a true lie? Is it truly a lie? Is it true that it's false? The wording of the question as well as the formatting of the Accusation are confusing for the average user. THIS REMAINS ONE OF THE MAJOR OBSTACLES TO GETTING SPONSORS.

    Answer: No
    Answer Confidence: 90 %
    Supporting Text:

    Answer: Don't Know
    Answer Confidence: 90 %
    Supporting Text:
    It is not a lie when the meaning of understanding in the context is given.

    Answer: No
    Answer Confidence: 75 %
    Supporting Text:
    Obviously there are individuals who don't understand how it works, but the question here is, as I understand it, whether humankind can understand it, and the answer is yes, they can.

    Answer: No
    Answer Confidence: 0 %
    Supporting Text:

    Answer:
    The deceit is that the lie is misleading.
    Answer Confidence: 100 %
    Supporting Text:

    Answer:
    The deceit is that the lie is misleading.
    Answer Confidence: 90 %
    Supporting Text:

    Answer:
    The deceit is that the lie is misleading.
    Answer Confidence: 90 %
    Supporting Text:

    Answer:
    The deceit is that the lie is misleading.
    Answer Confidence: 100 %
    Supporting Text:

    Answer:
    The deceit is that the lie is misleading.
    Answer Confidence: 100 %
    Supporting Text:
    I agree with the Plaintiff. That being said, THIS IS ONE OF THE MAJOR LINGUISTIC PROBLEMS OF THE PLATFORM. Is the Accusation a true lie? Is it truly a lie? Is it true that it's false? The wording of the question as well as the formatting of the Accusation are confusing for the average user. THIS REMAINS ONE OF THE MAJOR OBSTACLES TO GETTING SPONSORS.

    Answer:
    The deceit is that the lie is misleading.
    Answer Confidence: 90 %
    Supporting Text:
    To argue that something is impossible without qualification of terms is a type of deceit.

    Answer:
    The deceit is that the lie is manipulating.
    Answer Confidence: 90 %
    Supporting Text:
    LLMs are not properly controllable, that is the problem. Whether we understand them or not is beside the point.

    Answer:
    The deceit is that the lie is misleading.
    Answer Confidence: 80 %
    Supporting Text:

    Answer:
    The deceit is that the lie is misleading.
    Answer Confidence: 90 %
    Supporting Text:
    Not everyone is clued up on AI

    Answer: Yes
    Answer Confidence: 90 %
    Supporting Text:
    I believe a lot of people know better than to spread this lie about modern AI. They do it to make money or look smart because the average person does not have a clue about how these systems work. But the average person can understand how they work if they are informed and they think critically about it.

    Answer: Don't Know
    Answer Confidence: 90 %
    Supporting Text:

    Answer: Yes
    Answer Confidence: 90 %
    Supporting Text:
    I believe a lot of people know better than to spread this lie about modern AI. They do it to make money or look smart because the average person does not have a clue about how these systems work. But the average person can understand how they work if they are informed and they think critically about it.

    Answer: Yes
    Answer Confidence: 100 %
    Supporting Text:
    I believe a lot of people know better than to spread this lie about modern AI. They do it to make money or look smart because the average person does not have a clue about how these systems work. But the average person can understand how they work if they are informed and they think critically about it.

    Answer: Yes
    Answer Confidence: 90 %
    Supporting Text:
    I agree with the Plaintiff. This question is value-neutral and goes to the intent of the speaker of the Accusation. It works FAR better for this platform and requires that the juror offer some opinion as to whether the deceit or truth is intended by the speaker.

    Answer: Don't Know
    Answer Confidence: 90 %
    Supporting Text:
    I've listened to advocates whose credentials concerning understanding of AI are valid argue that nonetheless, it is not understood well enough to not be very wary of its potential to create harm.

    Answer: Don't Know
    Answer Confidence: 90 %
    Supporting Text:
    The motivation here is mysterious since LLMs are designed not to be tracked through every decision tree, understanding them in the sense of the causes Is problematic. We know this since they lie and cheat which was not part of the intention of the design but just appeared as a consequence of the complexity.

    Answer: Don't Know
    Answer Confidence: 80 %
    Supporting Text:
    Presumably the lie is intended.

    Answer: Yes
    Answer Confidence: 90 %
    Supporting Text:

    Answer:
    The motivation is to convince you that the lie is factually true.
    Answer Confidence: 90 %
    Supporting Text:
    At is based, this lie is trying to make you think something that is not factually true, is factually true.

    Answer:
    I'm not sure what the motivation is.
    Answer Confidence: 90 %
    Supporting Text:

    Answer:
    The motivation is to convince you that the lie is factually true.
    Answer Confidence: 90 %
    Supporting Text:
    At is based, this lie is trying to make you think something that is not factually true, is factually true.

    Answer:
    The motivation is to convince you that the lie is factually true.
    Answer Confidence: 100 %
    Supporting Text:

    Answer:
    The motivation is to convince you that the lie is factually true.
    Answer Confidence: 90 %
    Supporting Text:
    At is based, this lie is trying to make you think something that is not factually true, is factually true.

    Answer:
    The motivation is to persuade you to hate something or someone.
    Answer Confidence: 90 %
    Supporting Text:
    To hate the current pace and trajectory of AI development.

    Answer:
    Is the accusation a genuine lie?
    Answer Confidence: 90 %
    Supporting Text:
    If LLMs are fully understandable then responsibility can be assigned. It is in the makers interests that the accusation not be a lie but emergent behaviour is by its nature not understood by the components beneath, so either way the makers are irresponsible.

    Answer:
    I'm not sure what the motivation is.
    Answer Confidence: 95 %
    Supporting Text:
    I don't know why people would tell this lie, say that AI was not understandable. Perhaps they want to control it and be the ones to use AI. Or maybe they are frightened of it and wish to ban its use.

    Answer:
    The motivation is to be informative
    Answer Confidence: 90 %
    Supporting Text:

    Answer: Unacceptable
    Answer Confidence: 90 %
    Supporting Text:
    The media needs to change the story because it only damages the good that can come of modern AI used as a tool by everybody.

    Answer: Acceptable
    Answer Confidence: 90 %
    Supporting Text:
    In my honest opinion I feel like anything to do with AI is an acceptable topic. I feel that it's something that needs to be discussed more due to the fact that as the days go by and years go into years, the future of technology is going to rely on AI and therefore I feel like it should be an acceptable topic

    Answer: Don't Know
    Answer Confidence: 90 %
    Supporting Text:
    Most people seem to have accepted it sadly

    Answer: Unacceptable
    Answer Confidence: 100 %
    Supporting Text:

    Answer: Acceptable
    Answer Confidence: 100 %
    Supporting Text:
    Unfortunately, in a world where "if it bleeds, it leads" we see media outlets pushing negative narratives far more often solely to generate ad revenue.

    Answer: Unacceptable
    Answer Confidence: 90 %
    Supporting Text:
    Too general.

    Answer: Don't Know
    Answer Confidence: 90 %
    Supporting Text:
    The tool is limited usefulness in the task humanity has to grow up and behave sensibly,

    Answer: Don't Know
    Answer Confidence: 90 %
    Supporting Text:
    I don't know the reason behind the lie so I can't gauge its acceptability.

    Answer: Acceptable
    Answer Confidence: 90 %
    Supporting Text:

    Answer:
    This is factually untrue.
    Answer Confidence: 90 %
    Supporting Text:
    Modern AI can be understood as massive collections of very simple Perceptron computations.

    Answer:
    No label needed
    Answer Confidence: 90 %
    Supporting Text:

    Answer:
    No label needed
    Answer Confidence: 90 %
    Supporting Text:

    Answer:
    This is factually untrue.
    Answer Confidence: 90 %
    Supporting Text:
    AI can be managed and controlled if it’s regulated.

    Answer:
    The way that perceptrons work can be used very effectively to explain how AI works as well.
    Answer Confidence: 100 %
    Supporting Text:
    When discussing perceptrons though, we need a better description of their function and potential multi-state nature. The overwhelming number of people (myself included) think of computers functioning in a binary state. If there is another option to better explain AI computation, then a better description of how that option works would be helpful for everyone.

    Answer:
    This is factually untrue.
    Answer Confidence: 90 %
    Supporting Text:
    AI is too broad a topic to declare that it is generally impossible to understand.

    Answer:
    This is true, but misleading.
    Answer Confidence: 90 %
    Supporting Text:
    Understanding the architecture does not bestow meaning nor purpose. Humans using it without understanding will get into trouble.

    Answer:
    Some amongst us can understand generally how AI works, even of we cannot track every element of its functioning.
    Answer Confidence: 60 %
    Supporting Text:

    Answer:
    No label needed
    Answer Confidence: 90 %
    Supporting Text: