Elon Musk is quoted as saying in the summer of 2017 that AI presents “a fundamental existential risk for human civilisation”. Travel with me for a moment to the year 2026…
The integration of AI with our lives is pervasive in first world countries as so many companies have been racing with their technology solutions towards market share and commercial acclaim. With smart homes sensing our presence and adjusting the environmental controls, personalized medicine managed through wearables, and our vehicular travel being managed and coordinated for us, the common person can no longer distinguish between connected and unconnected living.
BREAKING NEWS: An elderly man in [any semi-urban area of the US] was found dead on Saturday afternoon after reports from his adult children that his digital presence had not been active for over twelve hours. After the automated vehicle from the local authority navigated through road hazards and downed trees near the home on Oak Lane, the accompanying Help-bot entered the home and found Mr. Smith unresponsive in the bedroom. He had expired at least eight hours before from asphyxiation as a result of a variety of converging incidents. The local authorities in collaboration with the national weather service and MegaHealthcare Services believe that high winds brought down heavy trees which interrupted CentralPower Co. utility services to the home in repeated, sporadic ways (i.e., a “brown-out”), which likely corrupdated the firmware in the homeowner’s smart-bed and environmental control (EvC) system. The bed which malfunctioned and adjusted to the maximum fully upright position along with the high-heat (nearly 100 deg) from the failed EvC system contributed to the airway obstruction that caused the man’s death as it bent the man in half and suffocated him. However, the primary contributor to his death was the failure of the smart-medical dosing machine that was coded incorrectly. Investigators say that this was a consequence of data error during upload of the prescription which transposed digits on the man’s birthday resulting in the incorrect medicine and dosage being administered. MegaHealthcare and CentralPower, while bearing no liability in the matter, express their condolences to the family and offer 100 NetCredits towards cremation services. Memorial services will be held Thursday and the family asks that contributions be made in Mr. Smith’s honor to the Robert Heinlein Park project in lieu of flowers.
No, I don’t actually believe that Artificial Intelligence will bring along doomsday for mankind; but, I do think that AI does present a great risk/threat as it will be greatly flawed as long as humans are designing, structuring and recording data. I have no doubt that humans will be able to develop wicked cool algorithms that can interpret data in new ways; and, that processes and technology will continue to advance at such a rate that we will have more data capacity and information analysis capability than we know what to do with (which is when the AI becomes sentient and begins to build its own capacity and capability in a self sustaining way – but that’s a topic for another conversation).
Let me get to my point…
In my many decades of data and information management I have NEVER seen a completely clean data set. And every AI project that I am close to right now has some level of “data cleanup” involved. Data cleanup – one of my most DESPISED terms – refers to a reactive effort to correct (even re-code or change) data in order to be used for query, analysis and even AI.
Truly valuable AI will need to rely on real-time or near-real-time data access to data that is accurate and trustworthy! The AI we develop and apply will always be only as good as the data to which we apply it (I know, duh).
SO…
Until we are capable of defining, developing and creating data sets that meet standards of data integrity and data quality that have the highest degree of accuracy and truthfulness, the AI results will be as flawed as the irrelevant junk mail that comes to my home every week.
