The machine learning is bringing many advantages in technology, but sometimes shows its limits: it is the case of Bixby, which is not yet available in English for the lack of data to be processed by the machine learning.
The problems that Samsung is having in preparation for the launch of Bixby , his voice assistant, would result from a lack of data. The machine learning system behind the software would not in fact sufficient data to be effective in English interactions, in contrast to the Korean language where Samsung seems to be able to achieve the desired results.
Human languages are particularly complex to deal with because of the many variables that distinguish them: differences in speakers’ voices, tones, inflections, accents, pronunciations, emphasis, sentence structure, word order, choice of vocabulary … To address all these aspects must have a base of data to be processed to have a picture as widely as possible of the situation; all data is then processed by a machine learning algorithm , which will extract the relevant information.
This means you can not enter data in a more or less manual, but you need to get to the starting block having already collected an amount of data sufficient to have a system that can address the use by users. To get out of what seems like a never-ending circle, Samsung has launched a beta of Bixby in which you can participate voluntarily, but it seems that will take some time before this initiative bear fruit.
If Samsung had initially announced that Bixby would be launched in June, now is not about any specific date for the voice assistant’s debut. With a loudspeaker connected in the works and Italian version that is not likely to get in a short time (if it ever will), Samsung seems to have some difficulty in dealing with the theme of the voice as a means of input .