Summing Up
Personal Predictive Analytics: Should We Be Careful What We Wish For?
The world of continuous monitoring of numerous sensors for machines and humans, limitless information storage capacity, and big data combined with rapid response logistics to detect and meet personal needs that we don't know we have can be exciting, efficient, and energizing. Or it can be fearsome, enervating, and subject to mistakes ranging from foolish to fatal. That's the sense provided by responses to this month's column.
There was enthusiasm for the potential of these capabilities. As Susan Kalyan put it, "Businesses are becoming so complex that integrating predictive analytics in decision making is not very far away… I think 'predictive analytics' is the 'Moneyball' for businesses." Kapil Kumar Sopory said that, "Prevention is better than cure, and this is what predictive analysis in effect helps us to achieve." Casandra Levine agreed, but introduced a cautionary note. The promise of what can be accomplished by predictive analytics in raising the quality of life and functionality for our society is astonishing, she wrote. "What is not clear yet is how those … qualitative factors will be mitigated (by) leaders who are short on ethics and morality…."
Others, most of whom assumed that the technologies would successfully be applied, were less sanguine about the results, expanding on Levine's concerns.
Karine Gantin asked, "If people are reminded over and over that some giant computers controlled by big corporations and institutions know better than they do about what they will be doing in (the) future and what they'll become … They might turn into poor, weak machines themselves …, abiding by what has been foreseen for them by algorithms." Gerald Nanninga added: "Predictive analytics could … (provide) such specialized, repetitive solutions that society will become polarized into user niches that can no longer properly interact… What a sad state of affairs." Philippe Gouamba cautioned that although we rejoice in being able to know the future, "Here is the problem: What happens when we see a future that we cannot affect? That is when predictive analytics become extremely uncomfortable."
Several responses sought to deal with the concerns. Arie Goldshlager commented that, "Predictive analytics can and do produce remarkable results," but they can also produce many false negatives and positives that have to be anticipated and proactively managed. Jeff Rutherford suggested that control in the hands of beneficiaries of the analytics will address needs for privacy. As he put it, "if I'm using a computer wearable such as the new iWatch, and I'm offered a service that will alert my doctor of any health abnormalities via the device…, I'm going to agree to that. My health will win out over privacy in that case."
We have learned to live with and benefit from predictive analytics when they are not applied to individuals. All forecasting is a case in point. When the application becomes personal, it raises many questions that will have to be addressed.
In this case, should we be careful what we wish for? What do you think?
Original Article
In 2002, the film Minority Report introduced many of us to the world of predictive analytics. In it, an innovative technology allows Washington, D.C. to go without a murder for six years by helping Tom Cruise, chief of the Precrime Unit, to identify, arrest, and prosecute killers before they commit their crimes.
This was a case of the movies catching up to the business world. At that time, predictive analytics had been applied to the continuing maintenance of everything from CAT scan machines produced by GE to elevators made by Otis. It enabled these firms to sell "up time" rather than just products, thanks to a number of sensors and the continuing remote surveillance of the performance of these products.
Predictive analysis applied to humans is now one of the hottest concepts to come along. It is being made possible by a system of customer loyalty programs, big data, and cloud computing that enables the continuous collection, storage, combination, and analysis of data about each of us from a number of disparate sources. Pretty exciting, no?
Some years ago, we heard the story about the GE maintenance engineer who, based on information from the firm's advanced monitoring and predictive analytics, visited one of his hospital accounts to repair a CAT scan machine that had not yet failed. As he was confronted by puzzled hospital administrators, the machine indeed stopped functioning. More recently, many of us have heard the story about the Target customer who was sent information about products of interest to pregnant women before she knew she was pregnant. Target's Big Data analysis of hers' and others' purchases, combined with related information, had placed her in a cohort with other women known to be pregnant.
Predictive analytics will be essential to the development of concepts such as 30-minute package delivery that companies like Amazon have been contemplating. For years, logistics have been managed by principles such as that of "postponement and speculation." The idea is that to approach the best match between supply and demand at a reasonable cost, a supplier has two basic choices. One is to delay (postpone) committing inventory to a particular supply point for as long as possible through such things as careful forecasting of demand, rapid manufacture, and fast transport. The other is to invest (speculate) in long but economical production batches, slow but economical transportation, and large amounts of inventory that ensure an in-stock position when an order is received.
An argument can be made that any forecast and inventory is based on predictive analytics. But in the past, these analytics were applied to data that described behaviors of large groups of decision-makers. By contrast, tomorrow's version of this technique will be based on the analysis of massive files of individual profiles, from which predictions will be built that establish stock levels needed to support 30-minute deliveries. Personalized logistics will take a lot more than just drones.
Predictive analytics have the potential to produce remarkable services and longer lives. But before we become too enamored with them, it's important to remember what happened to Tom Cruise in the movie. He is eventually accused on a precrime basis of murder, with only 36 hours to determine whether the charge is accurate and, if not, who implicated him wrongly.
How important are these concepts to our future? Is this a big deal or just another buzz term in business for the next several years? Are you ready for predictive analytics applied to you? If not, what are you going to do about it? What do you think?
Predictive analytics can and does produce remarkable results, but companies do need to anticipate and manage its negative side-effects:
Personalization, in particular, can and often does get too personal: "If you got it wrong once, it [Could] outweighed getting it right 10 times."
https://ariegoldshlager.wordpress.com/2012/06/23/can-personalization-get-too-personal/
And predictive analytics, in general, often produce many false negatives and positives that have to also be anticipated and proactively managed.
Arie.
That example will most likely be replicated across household appliances as well. But, if I'm using a computer wearable such as the new iWatch, and I'm offered a service that will alert my doctor of any health abnormalities via the device monitoring, I'm going to agree to that. My health will win out over privacy, in that case.
This white paper - The Internet of Everything is Now - Now What? - discusses the IoT in more depth - http://bit.ly/1FfCIeA
Jeff Rutherford commenting on behalf of IDG and KPMG
I don't need some polymorphic, double throw down algorithim to tell me who is going to use my product or not. Can it tell me who is ready to convert from the competion to my services? Does it count the relationship aspect? or the emotional tie? or satisfaction? or perceived value that changes to the customer every so often?
Maybe Vegas can use it to predict winners, (maybe they already do- its called the House) heck my broker could use that tool.
Maybe we can help to develop the state of art with some ideas.
Human beings are an interesting lot.
We rejoice in being able to know the future. We are masters of our destinies as we manipulate and redesign our various paths through life, for ourselves and for our children and for others.
Here is the problem: what happens when we see a future that we CANNOT affect? That is when predictive analytics become extremely uncomfortable. Messing with Mother Nature can have dyer consequences.
As analytics become more self-learning and intelligent, we raise the spectre of something very untoward occurring. Steven Hawking cites AI as one of the catastrophic risks affecting humanity.
For example, There are currently over 300 applications creating digital task assignments for knowledge workers. Daily, literally millions of work assignments are created, statused, reassigned and closed. Each of these digital sentence fragments represent lexemes of work. Utilizing blockchain technology, already being used by IBM for IoT proof of concept prototyping, these lexemes could form the basis of some pretty amazing predictive analytics. Think about:
1. Project plans assembled from past work.
2. Just in time assignments via resource deconfliction and automatic and always on leveling.
3. Risk alerts based on lexeme trending.
4. Benchmarking and best practices mining.
Further, utilizing the contract constructs introduced to the blockchain community by the Ethereum Project, the gig based economy can actually be a new form of unionism with gig workers defining the contracts they will work under (e.g., work for hire versus open source). Bottom up contracts could be a very powerful tool to close income inequality.
- The opening scene of the movie: the man arrested for the murder he was about to commit claims he wouldn't have done it in the end. What about freedom of choice and being able to change one's mind until the very last minute, implies here the screenplay? Seen from the movie's lens, predictive analysis totally denies that dimension, - might tomorrow deny that fundamental right to people in some totalitarian perspective. Using predictive analysis for business is one thing, but uses by public institutions will impact differently on a philosophical and political level.
- In Europe, we commonly relate the absence of progress during the Middle Ages and the subsequent recurring hardships on populations to the extreme rigidity of social structures of the time. If people are reminded over and over that some giant computers controlled by big corporations and institutions know better than they do about what they will be doing in next future and what they'll become in the long term for every aspect of their lives, - will those people (=our future "we") still be able in their immense majority to conduct their lives with positive initiatives for themselves and the society as a whole? They might turn into poor, weak machines themselves instead, abiding by what has been foreseen for them by algorithms. In my opinion, this would just mean a regressive society going backward, one that is unable to sustain itself properly.
- Now, these two preceding points must be kept in mind alongside with the motto that the worst-case scenario must be considered seriously enough as it will prove to be an insisting temptation for powers of all sorts at different levels. However, is pessimism always right? We all know that the 20th century witnessed different kinds of major crimes against humanity all partly driven by pseudo-scientific considerations. Nevertheless, we can't forget that positive change also happens nor can we underestimate economic factors when we try to envision the future of humankind. If both wealth and ability to conduct one's own life in positive ways are more equally spread, then we will be able to invent new paths for ourselves, - we will be able to deal with tomorrow's technologies to the best of humankind's abilities at the appropriate level: the individual and community level, - the most creative one. Here, empowerment is key and not at all an outfashioned concept, - on the contrary. But ther
e are conditions to be met for that. One thing is certain: we live in a quickly transitioning world, so predictions are not what they pretend to be, - they're only scenarios... and food for action.
I would prefer the term forecasting analytics... in forecasting we are aware of the uncertainties ahead. We look for early indications to monitor before taking the appropriate action at the right time. the mindset is one we accept the possibility of being wrong thus we continue to monitor the situation even after taking the necessary actions
And then there are the changes based on the situation. Take, for example, a simple activity of procuring and consuming a meal. My choices will be different for each of these situations:
? Going on a date
? Eating alone
? Entertaining the Boss
? Flush with cash after just getting paid
? End of the month when cash is tight
? Eating with children
As a brand, I would think it would be a lot easier to focus on owning one position and making it easy for those who drift into needing that position to find me rather than chasing millions of individuals as they move about through roles and situations. Hence, I think the potential for predictive analytics with individuals is limited.
Here's another problem I have with this approach which I will illustrate with an example. In the old days of listening to radio, I was exposed to many different sounds and broadened my music appreciation. Now, when I listen to Pandora, they have me listen to only one type of music. There is no variety, no broadening of interests, no learning, no growth. It's bad enough that our society has become so politically polarized because people can hide inside media which only spouts one extremist point of view. Predictive analytics could take that approach to our entire lives, providing such specialized, repetitive solutions, that society will become polarized into user niches that can no longer properly interact. There will be no variety, no growth, no cooperation, no learning. What a sad state of affairs.
This also enters into, and raises the question about (AI) artificial intelligence: when is to much personally acquired information, to much (and who will decide this)? Hence, if predictive analytics become the standard of operation, who will be in control of its predictive conscious; and will we as humans be in charge of our own? I stand by the notion, that: "All movements towards human capital advancements has its place in society; but knowing when the (its place) begins and ends - should be our/the primary concern). Simply stated, any intuitive system should not take the place of our ability as human beings to be self sufficient, and make decisions for ourselves (e.g., Predictive Analytics should be integrated with succinct measures of oversight/control; with the human capability to "opt in/out" - if one so desires).