Open I didn’t win because of one of the parameters was exploited and used as a cheat. A human needs to sleep, human thoughts are random and unlike human thoughts while sleeping, OpenAI was allowed to run its comparison program all night long giving Open AI an 8-hour times the number of days AI was operated during the time Open AI was training to fight the top gamer.
A new challenge is issued to OpenAI. The new parameters to train OpenAI would be the following:
Input variables mimicking the top player will be associated with the top-gamers daily life.
Every time the top-gamer takes a break, sleeps, goes to the store, random phones calls that all gamers hate, especially in that moment, every human trait of resting and not being around the machine must be programmed into the next OpenAI DOTA challenger. Other factors that weren’t programmed into OpenAI’s champion was outside interference. Interference that humans encounter while gaming.
Basically, OpenAI was allowed to train in an entirely and 100% sterile environment that humans do not have access to. That’s an exploit and it’s a bannable offense during any on-line video game contest between humans and AI.
Then we’ll see if the machine is better than a human.
Open I didn’t win because of one of the parameters was exploited and used as a cheat. A human needs to sleep, human thoughts are random and unlike human thoughts while sleeping, OpenAI was allowed to run its comparison program all night long giving Open AI an 8-hour times the number of days AI was operated during the time Open AI was training to fight the top gamer.
A new challenge is issued to OpenAI. The new parameters to train OpenAI would be the following:
Input variables mimicking the top player will be associated with the top-gamers daily life.
Every time the top-gamer takes a break, sleeps, goes to the store, random phones calls that all gamers hate, especially in that moment, every human trait of resting and not being around the machine must be programmed into the next OpenAI DOTA challenger. Other factors that weren’t programmed into OpenAI’s champion was outside interference. Interference that humans encounter while gaming.
Basically, OpenAI was allowed to train in an entirely and 100% sterile environment that humans do not have access to. That’s an exploit and it’s a bannable offense during any on-line video game contest between humans and AI.
In the new challenge, each side, AI and Human, gets to block one or two champions from each side before each event taking place between AI and Human begins.
Then we’ll see if the machine is better than a human.
LLM AIs don’t think, can’t think, and will never think. We aren’t even close to the vicinity of maybe having a clue on how to develop a “thinking” AI.
Oh, and their business model is essentially to lose a lot of fools’ money once the initial investors have cashed out. Turns out nobody is paying for the crap delivered by LLM AIs, they burn a ton of money to function, and there is no way in whcih they can ever become anything remotely intelligent.
Because AI isn’t inherently evil. It would have to be programmed that way. We’ve been killing each other in wars over land, gods, and money for millennia. Crime is rampant. AI doesn’t have those concepts. Humans are wants over needs. AI, unless programmed to the contrary, are needs over wants.
Still even more reason to keep AI out our lives, especially when competing. AI is like that swimming pool incident. Humans cant compete against a machine that does not sleep or feel anything.
But then again, if you could inject human algorythms into a bot, that would be the start of Skynet.
Let’s just hope that they will not implement a Frostpacker AI. Imagine the constant wordsalads 24 hours a day, 7 days a week, from multiple alts all at once, never ending.
The point that I am trying to make is, OpenAI defeated 5 of the top DOTA2 champs in two out of three matches.
OpenAI had a sterile environment in which to rapidly learn how to play DOTA 2. Wasnt a fair challenge to be honest.
Not until OpenAI is programmed using human learning variables will OpenAI actually be worthy of competing against players. OpenAI was basically set to God Mode.
If Musk is able to put his OpenAI program into a bot, what army could stand against Musk?
The question is, is the OpenAI program that defeated the DOTA 2 champions able to be uploaded into a real combat drone that would effectively learn its environment and then go after people?
The only difference between DOTA and real people is, probably nothing to the OpenAI program.
We are yet still far from AGI, but we would not probably even need them. The competition is already big between such form of intelligent forms.
I do not think building them as general kind of intelligences is efficient, specialization is key, then you tie all of them with a specialized AI overseer with a goal it was evolved to do. It is not like humans would ever function, and we can skip this part actually, because humans are already obsolete in many tasks needed for system like that to function.
but the more intriguing question, sorry to insist, in view of the dystopian scenario you propose is: could a future Musk ever be sure of being in control of such an army?
