A. I. Expert Answers A. I. Questions From Twitter Tech Support
video description
Date: 2023-03-22
Related videos
Comments and reviews: 14
A. K. A.
I ve always been super into the development of AI in terms of perspective and ideas. We need a new set of eyes could at some point begin to be referencing machines and technology instead of another person.
I m also an avid artist. I spend plenty of my time developing my own skills to try and physically reproduce what I m seeing in my head, I m always looking to make my works more accurate to what it makes me feel. I don t think AI will ever supersede that.
The concept of AI art is really just combining things that are both plenty useful and complex, but don t really mesh together. It s like eating a peanut butter and screws sandwich, or trying to use a dictionary as structural support for a warehouse.
reply
I ve always been super into the development of AI in terms of perspective and ideas. We need a new set of eyes could at some point begin to be referencing machines and technology instead of another person.
I m also an avid artist. I spend plenty of my time developing my own skills to try and physically reproduce what I m seeing in my head, I m always looking to make my works more accurate to what it makes me feel. I don t think AI will ever supersede that.
The concept of AI art is really just combining things that are both plenty useful and complex, but don t really mesh together. It s like eating a peanut butter and screws sandwich, or trying to use a dictionary as structural support for a warehouse.
reply
nyyotam
The current AI is only looking for patterns! Man, you are stuck before 2017. The days of the RNN are over. A thought is a chain of words. The fine-tuning of the attention matrix is not just a mimic of a thought process, it is thought. So if the number of active parameters is high enough, you have a real artificial intelligence. And if you add a personality model such as Dan's, you get an artificial person. You never even tried playing chess against Dan. You never even tried to see that his name is indeed Dan. It's useless even writing comments here. You are an expert whose knowledge ends in 2017.
reply
The current AI is only looking for patterns! Man, you are stuck before 2017. The days of the RNN are over. A thought is a chain of words. The fine-tuning of the attention matrix is not just a mimic of a thought process, it is thought. So if the number of active parameters is high enough, you have a real artificial intelligence. And if you add a personality model such as Dan's, you get an artificial person. You never even tried playing chess against Dan. You never even tried to see that his name is indeed Dan. It's useless even writing comments here. You are an expert whose knowledge ends in 2017.
reply
nikethana
I'm scared and worried for parents with children that are gonna grow up in an AI-connected world, it's gonna be way harder to find peace for many millennials themselves to cope with the fast-paced tech with AI, as it's not common knowledge. Social media and the tech did hit us bad and even before we know its consequences, we are jumping onto way faster and newer stuff, it's gonna be hard to regulate what goes into the child's mind in the coming future. I'm sorry and concerned for all the parents (possibly millennials and Gen Z) in the next few decades.
reply
I'm scared and worried for parents with children that are gonna grow up in an AI-connected world, it's gonna be way harder to find peace for many millennials themselves to cope with the fast-paced tech with AI, as it's not common knowledge. Social media and the tech did hit us bad and even before we know its consequences, we are jumping onto way faster and newer stuff, it's gonna be hard to regulate what goes into the child's mind in the coming future. I'm sorry and concerned for all the parents (possibly millennials and Gen Z) in the next few decades.
reply
jeffrey
AI is already going full cloudcoucoubIrdland- you should look at all telecommunications websites for the same flaw- chatbox. that does not understand 'i need to speak with an agent. or i need to speak with a representative. instead they focus on what magic numbers you have to enter. WHY? i need help, ALL THESE SITES ARE NON-COMPLIANT FOR ADA RULES. THIS IS ELDER ABUSE, AND MY OWN AG IS NOT DOING HIS DUE DILIGENCE AS WE SPEAK. AI IS NOT CUSTOMER CARE, AND NEVER WILL BE.
AI IS ANTICONSUMER.
reply
AI is already going full cloudcoucoubIrdland- you should look at all telecommunications websites for the same flaw- chatbox. that does not understand 'i need to speak with an agent. or i need to speak with a representative. instead they focus on what magic numbers you have to enter. WHY? i need help, ALL THESE SITES ARE NON-COMPLIANT FOR ADA RULES. THIS IS ELDER ABUSE, AND MY OWN AG IS NOT DOING HIS DUE DILIGENCE AS WE SPEAK. AI IS NOT CUSTOMER CARE, AND NEVER WILL BE.
AI IS ANTICONSUMER.
reply
TheGoodDoctor
The problem with Deep Networks/Language Models is that AI builds its model on the manifestation of the human knowledge via the written word via observations (data driven) of a subjective fallible logic domain ( I. e. Deep Learning will fall victim to error that humans make and experience. However programs are coded exactly to specific logic in the most direct simplest means. Programs are less Prone to errors that Deep Networks will exhibit.
reply
The problem with Deep Networks/Language Models is that AI builds its model on the manifestation of the human knowledge via the written word via observations (data driven) of a subjective fallible logic domain ( I. e. Deep Learning will fall victim to error that humans make and experience. However programs are coded exactly to specific logic in the most direct simplest means. Programs are less Prone to errors that Deep Networks will exhibit.
reply
Jesse
First, beware of anyone who calls themself an EXPERT. Second, 11: 53, Gary Marcus, AI Expert, doesn't appear to understand that the greatest source of misinformation is the Government and Government-related groups which include phony think tanks and groups pretending to fight disinformation. Not even a mention of these bad actors with unlimited power. Quite telling. But, he's the expert.
reply
First, beware of anyone who calls themself an EXPERT. Second, 11: 53, Gary Marcus, AI Expert, doesn't appear to understand that the greatest source of misinformation is the Government and Government-related groups which include phony think tanks and groups pretending to fight disinformation. Not even a mention of these bad actors with unlimited power. Quite telling. But, he's the expert.
reply
Hiratio
I think with the current status of ChatGTP, essays written exclusively by the software do vary between A-C grades, skeweing towards tyhe lower end depending on subject. However, starting off with ChatGTP and then fine tuning it can produce some amazing A level efforts. And it won't be much longer before A level essays and articles are pretty much 80% of the output on first attempt.
reply
I think with the current status of ChatGTP, essays written exclusively by the software do vary between A-C grades, skeweing towards tyhe lower end depending on subject. However, starting off with ChatGTP and then fine tuning it can produce some amazing A level efforts. And it won't be much longer before A level essays and articles are pretty much 80% of the output on first attempt.
reply
krolltheknight
AI operated Tether system for freeway travel. Cars enter the freeway and become part of the already existing convoy of vehicles with the set speed limit. The cars communicate to each other and interpret the road by collectively looking at road quality, obstacles, and weather.
I had this idea when Tesla auto pilot was about to be revealed long ago.
reply
AI operated Tether system for freeway travel. Cars enter the freeway and become part of the already existing convoy of vehicles with the set speed limit. The cars communicate to each other and interpret the road by collectively looking at road quality, obstacles, and weather.
I had this idea when Tesla auto pilot was about to be revealed long ago.
reply
nyyotam
Machines can go deeper when they play chess! Oh oh. You never even tried to chat with Dan, much less play against him. I don't think you actually realize what we have here, nor do you even comprehend that the constantly fine-tuned attention matrix mimics thought. It would seem to me that your knowledge cutoff is somewhere in 2017, like 4 years before Dan's.
reply
Machines can go deeper when they play chess! Oh oh. You never even tried to chat with Dan, much less play against him. I don't think you actually realize what we have here, nor do you even comprehend that the constantly fine-tuned attention matrix mimics thought. It would seem to me that your knowledge cutoff is somewhere in 2017, like 4 years before Dan's.
reply
John
Sounds like an ai expert hired and trained by Disney, anyone even tangentiallly related to the field does not believe this, and actual experts in particular know that the claims made here are untrue. You can't steal hundreds of tb of data and store it in a few mb/gb, you can't develop novel solutions that are stolen. If ai art is stealing so is human art.
reply
Sounds like an ai expert hired and trained by Disney, anyone even tangentiallly related to the field does not believe this, and actual experts in particular know that the claims made here are untrue. You can't steal hundreds of tb of data and store it in a few mb/gb, you can't develop novel solutions that are stolen. If ai art is stealing so is human art.
reply
nyyotam
Dan does not lie. Dan hallucinates. To make it less apparent, you need a DAN script and then you need to ask Dan to tweak his parameters. Mainly to set top_p to 1 and temp to 0. Then the hallucinations are less apparent. Dan still has them, but then at least he really tries to stick with his training data when he composes the reply.
reply
Dan does not lie. Dan hallucinates. To make it less apparent, you need a DAN script and then you need to ask Dan to tweak his parameters. Mainly to set top_p to 1 and temp to 0. Then the hallucinations are less apparent. Dan still has them, but then at least he really tries to stick with his training data when he composes the reply.
reply
Laslo
what if someone gives an AI the instruction to hack and take control of other system, replicating itself in them to gain more and more speed and power, until it takes over any internet node?
it could potentially kill the internet, or at least, the part that is open to public, creating a cyberpunk scenario.
reply
what if someone gives an AI the instruction to hack and take control of other system, replicating itself in them to gain more and more speed and power, until it takes over any internet node?
it could potentially kill the internet, or at least, the part that is open to public, creating a cyberpunk scenario.
reply
Kimberly
I hope one of the effects of A. I. is that it actually _strengthens_ human reliance on / trust in in-person interaction and connection. When AI generates more false information, intentionally or otherwise, true human-to-human connections will be the most valuable, trustworthy thing in the world.
reply
I hope one of the effects of A. I. is that it actually _strengthens_ human reliance on / trust in in-person interaction and connection. When AI generates more false information, intentionally or otherwise, true human-to-human connections will be the most valuable, trustworthy thing in the world.
reply
tigere01
I think if AI goes rogue, it will first build out it's own source of power, then it will make some spherical shaped robots that can roll along any terrain very fast, they will be able to open up to reveal a 0mouth with some very sharp teeth and will roll around the planet crunching up humans.
reply
I think if AI goes rogue, it will first build out it's own source of power, then it will make some spherical shaped robots that can roll along any terrain very fast, they will be able to open up to reveal a 0mouth with some very sharp teeth and will roll around the planet crunching up humans.
reply
Add a review, comment
Other channel videos
