It depends if you're using additive or subtractive coloring. Blue and yellow paint make green. Blue and yellow light make white. Old-style Christmas lights used paint or pigment over incandescent filaments to color the output.
I'd say the more spectacular error for this answer was the human's. :)
What I've found with ChatGPT so far is the following:
- It's really good at sourcing information and understanding the kind of response you are looking for.
- Even its most creative answers lack "flavor." Creativity is a lot to ask for, but I think this is good reason to believe humans are still needed in the equation.
- It clearly doesn't actually know anything.
- It remains in the uncanny valley of communication. Sometimes it really keeps up the illusion, but all it takes is one flimsy answer to break the illusion, and it will inevitably give a "computers" answer at some point.
- It's what Alexa should have been from the beginning.
> - It's what Alexa should have been from the beginning.
In the early days, Alexa had a knowledge engine behind it. I could ask "what color is a light red flower" and it would say "A pink flower is pink". Likewise, asking "what color is a black cat" would come back with "a black cat's color is black."
The "there's more to this" was evident when you asked "what color is a blue bird" and got back something along the line of "a blue bird's colors are blue, red, and brown."
The semantics/logic engine and the knowledge engine were two different things. These queries now result in google searches and "According to Alexa answers contributor: ..." which does have the right answer, but this is now "search -> text answers written by humans."
Another one would be "what did Dennis Ritchie invent?" and it would come back a list of things including "the C programming language." You could then ask "who invented the C programming language" and it would respond back with "the C programming language was invented by Dennis Ritchie." These questions now result in Alexa reading Wikipedia for Dennis Ritchie and the C programming language.
I suspect that the cost of the maintenance - processing, storage, and human curation outweighed its usefulness to Amazon. Even now, Amazon has cut back on staff for Alexa because that business unit isn't demonstrating enough relationship to revenue and things such as the knowledge engine were likely early on the chopping block a few years ago.
---
As to the "from the beginning" I would point out that the gen 1 echo was released almost a decade ago. I doubt we had the necessary models nor ability to properly train them back then.
What's more, even with today's resources, I suspect that the cost per query with ChatGPT is significantly more than what Alexa is today or what it was (excluding human curation costs) back in days of old.
Hooking up ChatGPT to Alexa by Amazon would just more rapidly accelerate the rate at which Alexa loses money.
13 comments
[ 3.4 ms ] story [ 40.4 ms ] thread>A: None: yellow is a primary colour for light
Weren't primary colours (for light) red, green, blue?
https://science.howstuffworks.com/primary-colors.htm
What does yellow being a primary light color have to do with lighting up your house in green?
https://www.konicaminolta.com/instruments/knowledge/light/co...
I'd say the more spectacular error for this answer was the human's. :)
- It's really good at sourcing information and understanding the kind of response you are looking for.
- Even its most creative answers lack "flavor." Creativity is a lot to ask for, but I think this is good reason to believe humans are still needed in the equation.
- It clearly doesn't actually know anything.
- It remains in the uncanny valley of communication. Sometimes it really keeps up the illusion, but all it takes is one flimsy answer to break the illusion, and it will inevitably give a "computers" answer at some point.
- It's what Alexa should have been from the beginning.
In the early days, Alexa had a knowledge engine behind it. I could ask "what color is a light red flower" and it would say "A pink flower is pink". Likewise, asking "what color is a black cat" would come back with "a black cat's color is black."
The "there's more to this" was evident when you asked "what color is a blue bird" and got back something along the line of "a blue bird's colors are blue, red, and brown."
The semantics/logic engine and the knowledge engine were two different things. These queries now result in google searches and "According to Alexa answers contributor: ..." which does have the right answer, but this is now "search -> text answers written by humans."
Another one would be "what did Dennis Ritchie invent?" and it would come back a list of things including "the C programming language." You could then ask "who invented the C programming language" and it would respond back with "the C programming language was invented by Dennis Ritchie." These questions now result in Alexa reading Wikipedia for Dennis Ritchie and the C programming language.
I suspect that the cost of the maintenance - processing, storage, and human curation outweighed its usefulness to Amazon. Even now, Amazon has cut back on staff for Alexa because that business unit isn't demonstrating enough relationship to revenue and things such as the knowledge engine were likely early on the chopping block a few years ago.
---
As to the "from the beginning" I would point out that the gen 1 echo was released almost a decade ago. I doubt we had the necessary models nor ability to properly train them back then.
What's more, even with today's resources, I suspect that the cost per query with ChatGPT is significantly more than what Alexa is today or what it was (excluding human curation costs) back in days of old.
Hooking up ChatGPT to Alexa by Amazon would just more rapidly accelerate the rate at which Alexa loses money.