‘EXPLAINABLE AI’ (XAI)
The human civilization has progressed on logic, reasoning and causality but with the onset of AI there is an arena where all the above is presumed but cannot be explained. Neural networks on which Artificial Intelligence edifice sustains still baffles researchers. The neural nets are ‘black boxes,’ after having been trained on data sets, the designers barely have an idea as to how the results are being generated. For this reason it is also ‘franken algos.’ This problem needs to be addressed as we move further and the road ahead thus of ‘explainable artificial intelligence,’AI.
Gartner Inc., forecasts that AI technologies by 2020 will become an integral part of nearly all new software product and service. According to recent Oxford & Yale University survey of over 350 AI researchers, machines better us in translating by 2024, writing high-school essays by 2026, driving a truck by 2027, working in retail by 2031, writing a book by 2049 and surgery by 2053. With AI taking over one task after another black box algos cannot be the future of mankind.
In March 2016, Microsoft Corp.’s chatbot generated racist tweets. It has to be pulled down and the company was forced to apologise. In June 2017, two AI chatbots as a part of Facebook Artificial Intelligence Research started talking in a language of their own presumably because English language rules did not suit them. Does our responsibility ends with writing an franken algo and get hopped on to a roller coaster or we decide the direction and fate of technology and mankind. Bots deciding & humans not even understanding is a frightful future. This is cognition upside down.
The sensible future according to David Gumming, Program Manager at the Darpa Information Innovation Office, would mean new machine-learning systems should have, “the ability to explain their rationale, characterize their strengths and weaknesses, and convey an understanding of how they will behave in the future.’ Companies and governments have to gravitate to Explainable AI also referred to as Transparent AI in the interest of our future.
AI ANTITHETICAL TO SAFE / UNKNOWN HUMAN FUTURE IS BETTER NOT CREATED.