Keeping in mind the adage on machine learning: data in, data out. I am curious whether there are neural network algorithms that could allow an algorithm to control and possess data on its own without the input from a programmer, or without human intervention.
I hope this Gif give some clarity...To automate we need Human Intervention or innovation or effort ..Automation obviously needs no intervention.
Ramesh Eluri Thanks for the answer. In your GIF, human intervention seems to have been removed from deep learning. Is that a closed universe data set now made up only of the data initially put in?
If yes, can deep learning be designed so that the data universe is open to future data without further human intervention?
1. Is that a closed universe data set now made up only of the data initially put in?
Yes, Deep learning systems require large amounts of data to return accurate results; accordingly, information is fed as huge data sets. When processing the data, artificial neural networks are able to classify data with the answers received from a series of binary true or false questions involving highly complex mathematical calculations. For example, a facial recognition program works by learning to detect and recognize edges and lines of faces, then more significant parts of the faces, and, finally, the overall representations of faces. Over time, the program trains itself, and the probability of correct answers increases. In this case, the facial recognition program will accurately identify faces with time.
2. can deep learning be designed so that the data universe is open to future data without further human intervention?
Yes, Deep learning is only in its infancy and, in the decades to come, will transform society. Self-driving cars are being tested worldwide; the complex layer of neural networks is being trained to determine objects to avoid, recognize traffic lights, and know when to adjust speed. Neural networks are becoming adept at forecasting everything from stock prices to the weather. Consider the value of digital assistants who can recommend when to sell shares or when to evacuate ahead of a hurricane. Deep learning applications will even save lives as they develop the ability to design evidence-based treatment plans for medical patients and help detect cancers early.
Eugene Veniaminovich Lutsenko What about self-coding AI? Could they be taught inadvertently to code so that they can come to control or possess data without the programmer's knowledge?
I think so, Yes. And this is a great danger. I think that if the opportunity exists, it will come true sooner or later. It's only a matter of time before that happens. It's like an Apple in Paradise. No matter how much you say, " Don't eat, don't eat!", it's still going to be eaten in the end
Aparna Sathya Murthy So, this is essentially a human choice whether to program AI to allow unfettered access to data. How about inadvertently allowing a self-programming AI to do so? Is that possible?
It is hard to avoid it. A program is a data structure, like a tree, or more general, a graph, and executing such a program is traversing it while interpreting the encountering objects as commands. One could prevent an AI system from secretly collecting own data and function collections by initializing its memory regularly.
Regards,
Joachim
Joachim Pimiskern So what I hear you say is that it would take active human oversight. Thanks for the insight.
The problem with any self-organised system is cohesion. That is, the more general is the system, the less cohesive it is. Consider expert systems. The value of an expert system is its limitation to a very restricted body of knowledge. Mycin is a great example. The inclusion of information outside of the domain of Mycin would necessarily reduce the cohesion of the database of facts upon which Mycin is based. Hence, the quality of answer given by Mycin for any particular case of symptoms would be reduced. It is hard to see how an abstract system can produce coherent rules for the interpretation of arbitrary phenomena.
- Badges
- Science topic