Post by eti800 on Feb 12, 2024 3:56:56 GMT -5
We have decided to write this post with the aim of shedding a little light on those of you who have concerns about artificial intelligence , wondering what it is supposed to be and what viable horizon of successes and failures is expected. To be fair, most of what we explain here is material taken from the work “Superintelligence” by Nick Bostrom. We highly recommend it.
Does artificial intelligence already exist?
Let's make a composition of where and where the development of artificial intelligence is. Let's take into account the following:
There is no social consensus (perhaps due to lack of knowledge or the complexity of the topic) when it comes to differentiating between AI, a simple computer program, and the automation of a traditional statistical process. This difficulty is often exploited to elevate any technology that Denmark Email List does not exceed the complexity of a spreadsheet to the category of intelligence.
It is not known how the brain works, nor is it known how intelligence arose as we conceive it. The organ that defines us as a species is a great unknown to us. What's more, there is no consensus on how many brains we really have. There are neurons in the heart and digestive system, but are they part of the brain? Do they influence conscious decision-making? etc.
Work is being done and partial achievements have been achieved in replicating cognitive mechanisms using electronic hardware and software , but very limited by current computing and storage capacity. In reality we know a lot about how the biological brain works (cause/effect), but very little about why it works in that particular way.
Artificial Intelligence vs. super intelligence
Artificial intelligence is confused with super intelligence, or what is the same, science with science fiction (at least for the moment).
Artificial intelligence could be defined as “one that can perform most human professions at least as well as a typical human being ,” while a superintelligence would be “any intellect that greatly exceeds the cognitive performance of humans.” in practically all areas of interest . ”
Artificial intelligence is the first step, the most realistic and for which we are working, super intelligence enters the field of metaphysics and will be a consequence, a leap into the abyss, with implications that are very difficult to predict, although today most are catastrophic and orbit around film scripts inspired by The Matrix or Terminator.
The long road to intelligent intelligence
What lines of work exist or have been evaluated to achieve advanced artificial intelligence comparable to human intelligence? Basically five:
From technology
1. Artificial intelligence: any general artificial intelligence system must be based on the ability to learn, the ability to manage uncertainty and the ability to create concepts . The first attempt was Alan Touring's children's machine, followed by the appearance of genetic algorithms that reproduce or direct the evolutionary and learning process that a brain experiences from birth. Currently there is also talk of “seminal AI”, which would be capable of improving its architecture.
2. Whole brain emulation: it is based on the creation of intelligent software from the scanning and modeling of a biological brain. There are already computer models of neurons and neuronal processes, but we insist again that the difficult thing is to find out the why and not the how, for this reason the objective in this line of work is reduced to achieving sufficient and essential properties that allow a intelligence to carry out a specific intellectual work. It is very likely that some of the other development paths will be successful before technology makes partial emulation of a biological brain possible.
Does artificial intelligence already exist?
Let's make a composition of where and where the development of artificial intelligence is. Let's take into account the following:
There is no social consensus (perhaps due to lack of knowledge or the complexity of the topic) when it comes to differentiating between AI, a simple computer program, and the automation of a traditional statistical process. This difficulty is often exploited to elevate any technology that Denmark Email List does not exceed the complexity of a spreadsheet to the category of intelligence.
It is not known how the brain works, nor is it known how intelligence arose as we conceive it. The organ that defines us as a species is a great unknown to us. What's more, there is no consensus on how many brains we really have. There are neurons in the heart and digestive system, but are they part of the brain? Do they influence conscious decision-making? etc.
Work is being done and partial achievements have been achieved in replicating cognitive mechanisms using electronic hardware and software , but very limited by current computing and storage capacity. In reality we know a lot about how the biological brain works (cause/effect), but very little about why it works in that particular way.
Artificial Intelligence vs. super intelligence
Artificial intelligence is confused with super intelligence, or what is the same, science with science fiction (at least for the moment).
Artificial intelligence could be defined as “one that can perform most human professions at least as well as a typical human being ,” while a superintelligence would be “any intellect that greatly exceeds the cognitive performance of humans.” in practically all areas of interest . ”
Artificial intelligence is the first step, the most realistic and for which we are working, super intelligence enters the field of metaphysics and will be a consequence, a leap into the abyss, with implications that are very difficult to predict, although today most are catastrophic and orbit around film scripts inspired by The Matrix or Terminator.
The long road to intelligent intelligence
What lines of work exist or have been evaluated to achieve advanced artificial intelligence comparable to human intelligence? Basically five:
From technology
1. Artificial intelligence: any general artificial intelligence system must be based on the ability to learn, the ability to manage uncertainty and the ability to create concepts . The first attempt was Alan Touring's children's machine, followed by the appearance of genetic algorithms that reproduce or direct the evolutionary and learning process that a brain experiences from birth. Currently there is also talk of “seminal AI”, which would be capable of improving its architecture.
2. Whole brain emulation: it is based on the creation of intelligent software from the scanning and modeling of a biological brain. There are already computer models of neurons and neuronal processes, but we insist again that the difficult thing is to find out the why and not the how, for this reason the objective in this line of work is reduced to achieving sufficient and essential properties that allow a intelligence to carry out a specific intellectual work. It is very likely that some of the other development paths will be successful before technology makes partial emulation of a biological brain possible.