Can small noise or signature help fix issues of dangers of natural language processing?

by Garima Khera, Deepak Singh, Raamesh Gowri Raghavan, and Sukant Khurana

 

Natural Language Processing (NLP) is a technical advancement which allows computers to analyse, understand and generate human language and speech. The speech can be easily copied using this technology, and can give birth to misuse of technology. Voice of any person can be used over telecommunication devices. Hence, a person’s voice may give threats to someone else while that person would be sitting at home watching sci-fi movie at the same time. Also certain programs that run on voice commands, use of NLP technology can also prove to be dangerous. We propose that, we in industry need to take steps so that in the court of law, an artificial speech can be discerned from human speech. While we want the technology to be good enough that future movies would have voice of long-dead artists but we do not want it would become a common weapon of misuse. While some issues of addressing how machines can discern if the voice is machine generated or human generated are addressable but at least we do not have at present a solution to propose for the speech to be discernible by humans. For machines to discern the speech, in all commercial products, a minor signature can be added to indicate the speech is machine generated.

References

· https://www.sas.com/en_us/insights/analytics/what-is-artificial-intelligence.html

· Shneiderman, B. (2007). Human Responsibility for Autonomous Agents. IEEE Intelligent Systems, Vol. 22 .

· F. Keith, R. M. William (2014) The Cambridge Handbook of Artificial Intelligence, Cambridge University Press

· S. L. A. Michael Anderson, “Machine Ethics: Creating an Ethical Intelligent Agent,” AI Magazine, vol. 28, no. 4. p. 15, 15-Dec-2007

· https://intelligence.org/files/EthicsofAI.pdf

· Association for the Advancement of Artificial Intelligence (www.aaai.org)