It’s not confused. That implies it understands what it wrote. The probability of g meaning grams must be higher than the force of gravity, even with the context. So grams was included in the generation of predicted text. Maybe it’s because g and G are both often used for gravity, where grams is always lower case.
Usually, g is used to represent acceleration due to gravity, whereas G is used to represent the gravitational constant in the calculation of the gravitational attraction between any two bodies.
G-force is used a lot. You are right if you look at it from what it’s being used for, but again LLMs are about statistical occurrences, and g/G is a good example of how the numbers in millions of data points may not match up to proper use.
It’s not confused. That implies it understands what it wrote. The probability of g meaning grams must be higher than the force of gravity, even with the context. So grams was included in the generation of predicted text. Maybe it’s because g and G are both often used for gravity, where grams is always lower case.
Usually, g is used to represent acceleration due to gravity, whereas G is used to represent the gravitational constant in the calculation of the gravitational attraction between any two bodies.
G-force is used a lot. You are right if you look at it from what it’s being used for, but again LLMs are about statistical occurrences, and g/G is a good example of how the numbers in millions of data points may not match up to proper use.
G-spot is also used, and I’m disappointed the AI didn’t hallucinate that into this response.
An AI that’s retrained/jailbroken might be more apt to go that direction.