- April 8, 2021
- Business
Not much doubt exists about the event “singularity” – the point at which AI surpasses human intelligence. The question is whether it will empower or displace human beings. That is, whether it will be benign. We possibly need to fear less about some super intelligent killer robots eliminating human race and fear more about Orwellian governments delivering dystopia. Dystopia comes from governments that have (1) access to large, very large, set of data about you, (2) the capability to analyze them all together and (3) supposed good intentions.
Nothing delivers a comfort of good intentions more than some narrow AI applications:
- Can a car move from point A to B without colliding with anything?
- Can a robot perform a cataract operation?
- Can a bank provide better loans basis your credit history?
- Can an insurance company provide better car insurance basis your driving record (acceleration, breaking, accidents, etc.)?
- Can an airline provide better seating basis your inflight behavior?
- Can one automate Whatsapp chat?
The most important use cases of Natural Language Processing are:
Narrow AI applications leads many to consider AI as an efficient and intelligent servant.
But the capability of AI is directly proportional to the available data – both in terms of width and depth. Depth refers to the number of consumer / people about whom there is data. Width refers to the richness or what all data that one has about a consumer / person. Firms go out of the way to expand both. Especially the latter. More one knows about a person, more power one has.
Sometimes for a firm, key data is available outside. Consider firms such as CIBIL that go about collecting your credit history across a host of platforms (banking, credit card, loans of all types, etc.). The credit scores decide whether you will get a loan and even is so, the rate. Firms, such as Zeotap, collect your data from a host of telecom, social media and other platforms to decide what kind of products you would buy today. The technologies are very real time and location & context specific. For e.g., if you are in McDonalds, it can entice you with an offer from KFC.
Large firms typically engage in strategic activities that expand the width and depth of available data. Consider Amazon that engages in a host of allied activities to eCommerce that expands the width of data available. For e.g., what you buy (amazon eComm), where your money is coming from (its bank, credit cards, pre-paid cards, store cards, or wallets and also loans), what you insure, what you watch (prime), what data is stored (AWS). Clearly, more comprehensive the data, more the ability to manipulate. Àmy Webb, NYU professor even has a funny acronym for Google, Microsoft, Amazon, Facebook, IBM, and Apple - G-MAFIA, for their ability to warp humanity with misuse of AI.
But nothing compares to what governments have and are wanting to have. Governments continue to expand the extent of data about a person and create laws that make private enterprises turn in data about private individuals. A motivated government has near unlimited budgets. Most projects have very noble intentions. But the road to hell is paved with good intentions.
Consider this: Chennai has the largest CCTV camera density per square kilometer. One cannot miss the red and blue flickering poles that mount a CCTV camera almost every 100 meters. The police state that the intention is to deter crime and review the data only if there is a report of a crime. But clearly, the utility of the data is larger. Traffic violations, jay walking, and more can be booked just through CCTV reviews. That capability is the first step to a surveillance society.
Governments already have access to your land or location records, education, health and financial data. It has the access to your social media data as well. Now add to it your movement data from CCTV cameras. What governments can do with it is truly mindboggling. Consider for e.g. the Chinese Social Credit Project.
Chinese Social Credit project aims to award each citizen points for good behavior and take away points for bad behavior. For e.g., from a facial recognition enabled traffic camera, you may be not just paying jay walking fines but also losing social credit points. What enhances social credit is not spitting, not jaywalking, not talking loudly in a metro, not fighting with one’s neighbor, not beating your children, eating healthy, not drinking in excess, working adequate hours, etc. etc. Not having adequate social credit points may mean, inability to book a flight, travel restrictions, house arrests, etc. One may say China is an exception to governance (or the lack of it) and one may argue that it may not happen in the US or Europe or India. But history has different tales to tell and those who ignore it are condemned to repeat it.
History reveals that all types of governments, democratic or communist, ruled by a king or elected representatives, black or white, have unleashed misery on hapless subjects when they could. There are just too few exception to this rule.
Stephen Hawking said, “the development of full artificial intelligence could spell the end of the human race.” His prophecy may have been from nightmares of fascists governments, with near unlimited budgets, creating monster AI with appearingly benign intent.
What can be done? The solutions are many. But here are three most important ones:
- Never provide to any one system of analysis all or comprehensive data. Either the width or the depth or both must be seriously curtailed.
- Create an autonomous and empowered data regulatory body – an authority that determines and regulates who should have what data, what they can do with it, how can they analyze it and how long they can keep it.
- Enshrine right to privacy of data under fundamental rights.
Until then, a rat should seriously consider whether it wants a kitten for a pet.
