subj
Pass the pipes :)
From Neil DeGrasse Tyson Official Site
"A 150,000-year-old rock formation near Mount Baigong in China lies one of archaeology's most perplexing mysteries: the Baigong Pipes. These strange, rust-colored metal pipes protrude from caves and stretch into a nearby saltwater lake, with some buried deep within the rock itself.
Their layout is strikingly geometric, too precise, many argue, to be a product of nature. What makes the phenomenon more intriguing is that local legends speak of "sky people" who descended from the heavens to build a great machine on the mountain long before a great flood reshaped the land.
Geologists have offered theories suggesting that the pipes could be fossilized tree roots or a natural formation caused by mineral crystallization and sedimentation. However, laboratory analysis of the pipe material revealed a high iron content and traces of rare metals not typically found in the region’s natural geology.
Furthermore, some of the pipes appear to be highly symmetrical and hollow, leading many researchers to question whether nature alone could have produced such structures especially at such a scale and depth.
What truly fuels the mystery is the suggested timeline. These pipes are embedded in rock believed to be around 150,000 years old, predating modern Homo sapiens.
If artificial, their existence could imply the presence of an unknown, technologically advanced civilization or visitors not of this Earth.
Whether the Baigong Pipes are remnants of an ancient industrial complex or a geological anomaly, they continue to defy conventional explanation and ignite the imaginations of scientists and theorists alike."
"A 150,000-year-old rock formation near Mount Baigong in China lies one of archaeology's most perplexing mysteries: the Baigong Pipes. These strange, rust-colored metal pipes protrude from caves and stretch into a nearby saltwater lake, with some buried deep within the rock itself.
Their layout is strikingly geometric, too precise, many argue, to be a product of nature. What makes the phenomenon more intriguing is that local legends speak of "sky people" who descended from the heavens to build a great machine on the mountain long before a great flood reshaped the land.
Geologists have offered theories suggesting that the pipes could be fossilized tree roots or a natural formation caused by mineral crystallization and sedimentation. However, laboratory analysis of the pipe material revealed a high iron content and traces of rare metals not typically found in the region’s natural geology.
Furthermore, some of the pipes appear to be highly symmetrical and hollow, leading many researchers to question whether nature alone could have produced such structures especially at such a scale and depth.
What truly fuels the mystery is the suggested timeline. These pipes are embedded in rock believed to be around 150,000 years old, predating modern Homo sapiens.
If artificial, their existence could imply the presence of an unknown, technologically advanced civilization or visitors not of this Earth.
Whether the Baigong Pipes are remnants of an ancient industrial complex or a geological anomaly, they continue to defy conventional explanation and ignite the imaginations of scientists and theorists alike."
subj
Comparing Oranges to Oranges
Confidently wrong, round two.
Asked ChatGPT to compare two orange circles. Without hesitation, it told me:
👉 “They’re the same size.”
But… they’re clearly not.
What happened? My guess: it’s seen this setup in its training data as the Ebbinghaus illusion (where surrounding circles trick your brain). So it confidently inferred that’s what I must be asking about, even though this time, the circles are obviously different.
This is a great reminder:
• LLMs don’t truly “see” or “understand” images.
• They rely on patterns in training data.
• And when they guess wrong, they do so with confidence.
Asked ChatGPT to compare two orange circles. Without hesitation, it told me:
👉 “They’re the same size.”
But… they’re clearly not.
What happened? My guess: it’s seen this setup in its training data as the Ebbinghaus illusion (where surrounding circles trick your brain). So it confidently inferred that’s what I must be asking about, even though this time, the circles are obviously different.
This is a great reminder:
• LLMs don’t truly “see” or “understand” images.
• They rely on patterns in training data.
• And when they guess wrong, they do so with confidence.
express, not stress