..and, when you have real expertise in a field - any field - and question an LLM chatbot about anything in that field, you realize really quickly how non-functional the tech is. It will argue and talk BS to justify its wrong answers, but: what is the use case for tech that does not really work?
November 9, 2025 at 1:38 AM
..and, when you have real expertise in a field - any field - and question an LLM chatbot about anything in that field, you realize really quickly how non-functional the tech is. It will argue and talk BS to justify its wrong answers, but: what is the use case for tech that does not really work?
Don't.
It´s buggy, inconsistent, makes everything harder to use due to nonsensical UI decisions. A complete mess. Shockingly bad, really. Major lag in the most basic operations, like scrolling a Finder window in icon view. Audio dropouts when going full screen in the Music app. Super buggy. 🤢
It´s buggy, inconsistent, makes everything harder to use due to nonsensical UI decisions. A complete mess. Shockingly bad, really. Major lag in the most basic operations, like scrolling a Finder window in icon view. Audio dropouts when going full screen in the Music app. Super buggy. 🤢
September 16, 2025 at 2:43 AM
Don't.
It´s buggy, inconsistent, makes everything harder to use due to nonsensical UI decisions. A complete mess. Shockingly bad, really. Major lag in the most basic operations, like scrolling a Finder window in icon view. Audio dropouts when going full screen in the Music app. Super buggy. 🤢
It´s buggy, inconsistent, makes everything harder to use due to nonsensical UI decisions. A complete mess. Shockingly bad, really. Major lag in the most basic operations, like scrolling a Finder window in icon view. Audio dropouts when going full screen in the Music app. Super buggy. 🤢
This is the first time I upgraded macOS and thought 'this was written by idiots'. Small random symbols in menus that make everything harder to read. Insane lag in displaying icons in Finder windows. Buttons as tabs. Avoidable transparency issues. Spotlight UI is a complete mess. Inconsistencies. ...
September 16, 2025 at 2:29 AM
This is the first time I upgraded macOS and thought 'this was written by idiots'. Small random symbols in menus that make everything harder to read. Insane lag in displaying icons in Finder windows. Buttons as tabs. Avoidable transparency issues. Spotlight UI is a complete mess. Inconsistencies. ...
testing, testing:
August 11, 2025 at 5:31 PM
testing, testing:
ChatGPT, latest release from today:
August 8, 2025 at 5:10 PM
ChatGPT, latest release from today:
Got the correct response for blueberry, but:
August 8, 2025 at 5:00 PM
Got the correct response for blueberry, but:
Or - shifting the conversation away from prejudiced stereotypes of persons or groups of people:
Truth: behavior of currently available tech, with understanding of the technical underpinnings, limitations, and real world experience.
Hype: hypothetical future tech, speculation based on conjecture.
Truth: behavior of currently available tech, with understanding of the technical underpinnings, limitations, and real world experience.
Hype: hypothetical future tech, speculation based on conjecture.
August 8, 2025 at 4:22 PM
Or - shifting the conversation away from prejudiced stereotypes of persons or groups of people:
Truth: behavior of currently available tech, with understanding of the technical underpinnings, limitations, and real world experience.
Hype: hypothetical future tech, speculation based on conjecture.
Truth: behavior of currently available tech, with understanding of the technical underpinnings, limitations, and real world experience.
Hype: hypothetical future tech, speculation based on conjecture.
The fact that you try to steer the conversation to personal attacks against me is the best indication that I have made valid factual points.
I have provided links to content from a premier tech site, a scientific study, and quotes from an LLM itself.
Truth is not 'anti-hype'. It's truth.
I have provided links to content from a premier tech site, a scientific study, and quotes from an LLM itself.
Truth is not 'anti-hype'. It's truth.
August 7, 2025 at 2:38 PM
The fact that you try to steer the conversation to personal attacks against me is the best indication that I have made valid factual points.
I have provided links to content from a premier tech site, a scientific study, and quotes from an LLM itself.
Truth is not 'anti-hype'. It's truth.
I have provided links to content from a premier tech site, a scientific study, and quotes from an LLM itself.
Truth is not 'anti-hype'. It's truth.
Good explanation:
'LLMs like GPT are pattern-matchers. They don’t know what words mean—they learn how words statistically relate to other words. Given a prompt, they predict the most likely continuation. This can produce breathtakingly fluent results, but it’s not the same as understanding.'
'LLMs like GPT are pattern-matchers. They don’t know what words mean—they learn how words statistically relate to other words. Given a prompt, they predict the most likely continuation. This can produce breathtakingly fluent results, but it’s not the same as understanding.'
August 6, 2025 at 6:40 PM
Good explanation:
'LLMs like GPT are pattern-matchers. They don’t know what words mean—they learn how words statistically relate to other words. Given a prompt, they predict the most likely continuation. This can produce breathtakingly fluent results, but it’s not the same as understanding.'
'LLMs like GPT are pattern-matchers. They don’t know what words mean—they learn how words statistically relate to other words. Given a prompt, they predict the most likely continuation. This can produce breathtakingly fluent results, but it’s not the same as understanding.'
Are you programmed so that your responses appear to be human, to intentionally make you more relatable to humans?
ChatGPT:
'Yes. ...'
ChatGPT:
'Yes. ...'
August 6, 2025 at 5:05 PM
Are you programmed so that your responses appear to be human, to intentionally make you more relatable to humans?
ChatGPT:
'Yes. ...'
ChatGPT:
'Yes. ...'
Would you describe yourself as intelligent?
Chat GPT:
'...I’m not truly intelligent in the human sense. I don’t have consciousness, self-awareness, emotions, or the ability to understand experience like humans do. I also can’t learn from past interactions..'
LLM said it, so it must be true.
Chat GPT:
'...I’m not truly intelligent in the human sense. I don’t have consciousness, self-awareness, emotions, or the ability to understand experience like humans do. I also can’t learn from past interactions..'
LLM said it, so it must be true.
August 6, 2025 at 4:47 PM
Would you describe yourself as intelligent?
Chat GPT:
'...I’m not truly intelligent in the human sense. I don’t have consciousness, self-awareness, emotions, or the ability to understand experience like humans do. I also can’t learn from past interactions..'
LLM said it, so it must be true.
Chat GPT:
'...I’m not truly intelligent in the human sense. I don’t have consciousness, self-awareness, emotions, or the ability to understand experience like humans do. I also can’t learn from past interactions..'
LLM said it, so it must be true.
Chat GPT says:
'My responses come from patterns in the large dataset I was trained on, which includes text from books, articles, websites, and other sources, up until my knowledge cutoff (which is in 2023).'
LLM = assembly of PATTERNS OF WORDS based on a statistical dataset.
'My responses come from patterns in the large dataset I was trained on, which includes text from books, articles, websites, and other sources, up until my knowledge cutoff (which is in 2023).'
LLM = assembly of PATTERNS OF WORDS based on a statistical dataset.
August 6, 2025 at 4:39 PM
Chat GPT says:
'My responses come from patterns in the large dataset I was trained on, which includes text from books, articles, websites, and other sources, up until my knowledge cutoff (which is in 2023).'
LLM = assembly of PATTERNS OF WORDS based on a statistical dataset.
'My responses come from patterns in the large dataset I was trained on, which includes text from books, articles, websites, and other sources, up until my knowledge cutoff (which is in 2023).'
LLM = assembly of PATTERNS OF WORDS based on a statistical dataset.
Emerging intelligence - the 'thinking' that you keep referring to, does not exist.
There. Is. No. Artificial. Intelligence.
Sure, there may be AI in a few decades, but right now, all there is is hype.
And - see Are Technica article above - increasingly low confidence among the tech enthusiasts.
There. Is. No. Artificial. Intelligence.
Sure, there may be AI in a few decades, but right now, all there is is hype.
And - see Are Technica article above - increasingly low confidence among the tech enthusiasts.
August 6, 2025 at 4:03 PM
Emerging intelligence - the 'thinking' that you keep referring to, does not exist.
There. Is. No. Artificial. Intelligence.
Sure, there may be AI in a few decades, but right now, all there is is hype.
And - see Are Technica article above - increasingly low confidence among the tech enthusiasts.
There. Is. No. Artificial. Intelligence.
Sure, there may be AI in a few decades, but right now, all there is is hype.
And - see Are Technica article above - increasingly low confidence among the tech enthusiasts.
arxiv.org/pdf/2506.11135
This paper from this year look into LLM emergence.
It differentiates between 'emergent properties' - being able to identify the outline of an object in an image background and making assumptions about what is behind the object - vs. 'emergent intelligence'.
This paper from this year look into LLM emergence.
It differentiates between 'emergent properties' - being able to identify the outline of an object in an image background and making assumptions about what is behind the object - vs. 'emergent intelligence'.
arxiv.org
August 6, 2025 at 3:59 PM
arxiv.org/pdf/2506.11135
This paper from this year look into LLM emergence.
It differentiates between 'emergent properties' - being able to identify the outline of an object in an image background and making assumptions about what is behind the object - vs. 'emergent intelligence'.
This paper from this year look into LLM emergence.
It differentiates between 'emergent properties' - being able to identify the outline of an object in an image background and making assumptions about what is behind the object - vs. 'emergent intelligence'.
The only place where magical thinking comes it is the assumption that a statistics program could have 'emergent properties' that are not explicit in its code, and are unexpected to the people who wrote the code.
An LLM giving different answers to the same question today and tomorrow is faulty code.
An LLM giving different answers to the same question today and tomorrow is faulty code.
August 6, 2025 at 3:27 PM
The only place where magical thinking comes it is the assumption that a statistics program could have 'emergent properties' that are not explicit in its code, and are unexpected to the people who wrote the code.
An LLM giving different answers to the same question today and tomorrow is faulty code.
An LLM giving different answers to the same question today and tomorrow is faulty code.
Or, could it be that it is an entirely understandable human trait to project human properties on tech?
There are no neural networks comparable to a human brain.
When a program, especially a complex program, does not do what the programmer expected, it is called an unexpected error. Hype ≠ Reality.
There are no neural networks comparable to a human brain.
When a program, especially a complex program, does not do what the programmer expected, it is called an unexpected error. Hype ≠ Reality.
August 6, 2025 at 3:17 PM
Or, could it be that it is an entirely understandable human trait to project human properties on tech?
There are no neural networks comparable to a human brain.
When a program, especially a complex program, does not do what the programmer expected, it is called an unexpected error. Hype ≠ Reality.
There are no neural networks comparable to a human brain.
When a program, especially a complex program, does not do what the programmer expected, it is called an unexpected error. Hype ≠ Reality.
When you open a text editor and type
I CAN THINK
and then print out the page and look at it, that does not mean that your printer suddenly acquired the the capability to think, and informed you about it.
I CAN THINK
and then print out the page and look at it, that does not mean that your printer suddenly acquired the the capability to think, and informed you about it.
August 5, 2025 at 5:44 PM
When you open a text editor and type
I CAN THINK
and then print out the page and look at it, that does not mean that your printer suddenly acquired the the capability to think, and informed you about it.
I CAN THINK
and then print out the page and look at it, that does not mean that your printer suddenly acquired the the capability to think, and informed you about it.
Please kindly educate me what these 'emergent properties' are supposed to be, and how they work on a technological level.
August 5, 2025 at 5:25 PM
Please kindly educate me what these 'emergent properties' are supposed to be, and how they work on a technological level.
'Artificial Intelligence' is a marketing term for a statistics engine. Phrases like 'think more deeply' are advertising.
An LLM is fundamentally not different to a spreadsheet program. It can do some basic calculations, but it cannot 'think' in any shape or form. Spreadsheet: 1+1=2, LLM 1+1=3.4
An LLM is fundamentally not different to a spreadsheet program. It can do some basic calculations, but it cannot 'think' in any shape or form. Spreadsheet: 1+1=2, LLM 1+1=3.4
August 4, 2025 at 10:13 PM
'Artificial Intelligence' is a marketing term for a statistics engine. Phrases like 'think more deeply' are advertising.
An LLM is fundamentally not different to a spreadsheet program. It can do some basic calculations, but it cannot 'think' in any shape or form. Spreadsheet: 1+1=2, LLM 1+1=3.4
An LLM is fundamentally not different to a spreadsheet program. It can do some basic calculations, but it cannot 'think' in any shape or form. Spreadsheet: 1+1=2, LLM 1+1=3.4
When you say 'think through' it suggests a fundamental misunderstanding of the tech. It is not actually 'Artificial Intelligence', it is a statistics program that tries to associate words, and outputs random word sequences that meet statistic criteria. Some make sense, some don't.
August 4, 2025 at 5:38 PM
When you say 'think through' it suggests a fundamental misunderstanding of the tech. It is not actually 'Artificial Intelligence', it is a statistics program that tries to associate words, and outputs random word sequences that meet statistic criteria. Some make sense, some don't.