Artificial intelligence is not something new, in fact artificial intelligence has been researched as far back as the early 20th century. However, the most notable example would have to be Alan Turing, with his paper ‘Computing Machinery and Intelligence’ from 1950. In this paper Turing discusses many intricate details pertaining to the now burgeoning field of artificial intelligence, which then was only in its infancy – more or less. This paper is what led to what is now known as the Turing Test – which is a test that judges a machine’s “intelligence”. Of course Turing did not refer to it as the Turing Test, for him it was simply a computer’s participation in the ‘imitation game’ (yes, like the movie), where the computer (A) would attempt to make an interrogator (C) think that A is in fact B, which is another person. Hence the name the ‘imitation game’. A can lie, but it is generally accepted that B will tell the truth about themselves. The game is played through computer screens. There is much, much more to the paper, however the basic premise for this article is that the Turing Test is designed to evaluate the effectiveness of artificial intelligence – i.e. “Can machines think?”.
These days artificial intelligence seems to be leaning more toward the ability of computers to “learn”. The process might go something like this – an algorithm is created → relevant data is inputted → it is tested → it is given feedback by humans → the algorithm adjusts according to human feedback. Obviously this is a simplified version of one type of artificial intelligence, but it is one way in which machines can learn from humans. This is commonly referred to as ‘machine learning’. The advancement of such artificial intelligence systems will lead, and are in fact leading to self-learning, where it is not necessary to have constant human feedback. These systems, (such as neural networks) can be (and are proposed to be) used in the recovery of data.
Igor Sestanj and David Edwards propose that machine learning can be used in the recovery of flash and NAND devices such as USB drives and SD cards. They liken the process to that of the human neural system (where neurons accept information via dendrites, process the information and make “decisions”, which lead to an output), and that because many manufacturers use predictable XOR patterns for writing data, firmware corruption can be reversed by replicating the manufacturer’s XOR pattern – revealing the original raw data, and hence be able to be recovered. Here is a link to the paper itself for those that would like to read more – the future of data recovery is very exciting indeed!
As always reader participation is not just welcomed, but encouraged! If you have any suggestions, corrections or anything in between, feel free to leave a comment. If the content of this blog is gobbledygook and you are in need of data recovery or other IT services, head on over to our Web Shop (currently undergoing maintenance). Want a good laugh? Check out our blog created entirely by artificial intelligence (AI)
We’ve done the research, so you don’t have to!
Thanks for reading!