Cortical Labs are the ones who pulled this off. They already have biological computers running on 800,000 lab-grown neurons available for ~$35,000 (just going on what a quick Google search told me) and are planning to open up a cloud computing service with its own API soon.
This makes me feel uneasy. Imagine if reincarnation were a thing and you get brought back into this world, and your purpose is to learn how to play DOOM.
Personally my worry really isn’t reincarnation, there’s no reason to believe that that’s true. But if these are fundamentally the same neurons that make up our brains, then how much do you need to put together before they acquire some form of “sentience”? Does a clump of 800,000 human neurons experience pain, sadness, a sense of self? Where is the line between an emotionless biocomputer and torturing a living organism for its entire lifespan?
Despite the fact that I really hate “AI”, that question was of course already sort of relevant for the latest AI models, even though we can generally conclude that they’re not there yet at all. But real neurons are different, we know what they’re capable of. How many do you need before a clump of neurons has rights?
Large language models are not intelligent. They are predictive text applications with massive dictionaries of circumstantial sentence structures to choose from. Nothing more. They do not feel and do not think for themselves. The only time they do anything is when the API calls them to produce more text with an updated context string.
Sure, but is the full human brain the minimum set necessary?
Sentience/sapience is probably an emergent property of a set of neurons needing to coordinate, plan, predict the future and oneself in relation to it.
I suspect that AI is capable of sentience with sufficient complexity and training, but it’s not there yet. I also suspect we’ll be well past the point where it is there before we realize it is, but not until we make some kind of fundamental change in how we do it - we know human level intelligence is possible in the volume and power consumption of, well, a brain so we’re orders of magnitude off of efficiency limits.
It’s estimated that mice have 70 million to 100 million neurons in their brains. They are capable of feeling pain and have social hierarchy. They also experience emotions like fear, pleasure, and anxiety. (We use them in pharmacology models of many mental illnesses.)
Have you ever heard the phrase, “the neurons that fire together, wire together” ? Our neurons are in a constant feedback loop with the environment we experience. Our experiences shape how our neurons make interconnected networks, which then impacts how we behave upon the environment.
If those neurons connected to the computer chip only ever experience playing the game “DOOM,” how would they know about anything else? How could they know about pain without having limbs to innervate and experience the pain with? How could they have a social hierarchy without others to interact with? We may as well be god to those neurons on the PC chip, because we are controlling the entire world they have access to.
What I find sad is that our society is ok with hooking living cells up to a computer to make smarter computers, but has a problem with ethically harvesting stem cells to be used to treat diseases.
Cortical Labs are the ones who pulled this off. They already have biological computers running on 800,000 lab-grown neurons available for ~$35,000 (just going on what a quick Google search told me) and are planning to open up a cloud computing service with its own API soon.
This makes me feel uneasy. Imagine if reincarnation were a thing and you get brought back into this world, and your purpose is to learn how to play DOOM.
aw sweet, man made horrors beyond my comprehension 😍
There’s another bunch of guys who are trying to do the same thing with rat neurons on the cheap using Gatorade as a growth medium.
https://youtu.be/bEXefdbQDjw
Personally my worry really isn’t reincarnation, there’s no reason to believe that that’s true. But if these are fundamentally the same neurons that make up our brains, then how much do you need to put together before they acquire some form of “sentience”? Does a clump of 800,000 human neurons experience pain, sadness, a sense of self? Where is the line between an emotionless biocomputer and torturing a living organism for its entire lifespan?
Despite the fact that I really hate “AI”, that question was of course already sort of relevant for the latest AI models, even though we can generally conclude that they’re not there yet at all. But real neurons are different, we know what they’re capable of. How many do you need before a clump of neurons has rights?
Large language models are not intelligent. They are predictive text applications with massive dictionaries of circumstantial sentence structures to choose from. Nothing more. They do not feel and do not think for themselves. The only time they do anything is when the API calls them to produce more text with an updated context string.
I don’t know how many neurons are in a human brain, but if you made an artificial human brain, could it have consciousness?
It has to be a full fetus with a heartbeat to have rights. /s In all seriousness, the human brain is estimated to have 86 billion neurons.
Sure, but is the full human brain the minimum set necessary?
Sentience/sapience is probably an emergent property of a set of neurons needing to coordinate, plan, predict the future and oneself in relation to it.
I suspect that AI is capable of sentience with sufficient complexity and training, but it’s not there yet. I also suspect we’ll be well past the point where it is there before we realize it is, but not until we make some kind of fundamental change in how we do it - we know human level intelligence is possible in the volume and power consumption of, well, a brain so we’re orders of magnitude off of efficiency limits.
It’s estimated that mice have 70 million to 100 million neurons in their brains. They are capable of feeling pain and have social hierarchy. They also experience emotions like fear, pleasure, and anxiety. (We use them in pharmacology models of many mental illnesses.)
Have you ever heard the phrase, “the neurons that fire together, wire together” ? Our neurons are in a constant feedback loop with the environment we experience. Our experiences shape how our neurons make interconnected networks, which then impacts how we behave upon the environment.
If those neurons connected to the computer chip only ever experience playing the game “DOOM,” how would they know about anything else? How could they know about pain without having limbs to innervate and experience the pain with? How could they have a social hierarchy without others to interact with? We may as well be god to those neurons on the PC chip, because we are controlling the entire world they have access to.
What I find sad is that our society is ok with hooking living cells up to a computer to make smarter computers, but has a problem with ethically harvesting stem cells to be used to treat diseases.
It’s because the stem cells somehow threatened the religious hegemony.
“Do lab grown neurons have a soul?”
I would say consciousness is required for that, so no.
People used to say animals were not concious.
Recent science suggest that some animals have what humans would consider to be language. This is a slippery slipe.
A lot of religious people still say that.