Those who said that sufficiently advanced technology is indistinguishable from magic were not wrong at all, and between MIT researchers and Hogwarts students the gap is narrowing quite a bit, thanks to AI developments.
The results are already astonishingly accurate and show a terrifying example of what a more sophisticated AI will come up with from data analysis.
In a study published on arXiv this week, the Massachusetts institute's team described training a computational network to analyze small voice clips and compare them with different biometric characteristics of the clips' authors. On the first try he has already “guessed” the owners of the items at a rate above chance, and he becomes more and more skilled.
The research brings with it a wide load of debate on positive (security and espionage, ability to identify subjects also from voice maps) and negative (mainly related to ethics and privacy, and they are not trivial aspects) of this research.
“Although this is a simple academic research, I think it was important to offer a series of ethical considerations related to the potential of this algorithm already on the sidelines of this study,” they explain.
The wonder remains despite the concerns, and with it the certainty that this instrument will also become part of a military and/or civilian "kit" in the near future.