Age of computers and Data, where do Humans Stand

One thing that computers can't create is knowledge....and the real question is the human knowledge and discovery at a place where the incremental effort and value is just re-gurgitation, maybe complex combinations that appear unique but are patterns and insights about existing knowledge.

An increase in computing power and understanding of pattern recognition in human communication, be it facial expressions or speech recognition or language translation, is making emotionless machines appear cognitive. Pleasure, pain, memory and experience, combined with the contexts (surroundings, circumstances) in which all those occurred, and intensity of each experience,  along with the DNA of the individual that makes the absorption of the experience unique to the subject, eventually mold the decision making into a personal trait. In short the inputs that go into a human machine with abstractions of data signals is almost infinite. Although many particular behaviors of humans have become more predictable, at individual level maybe it is hard to guess which way a particular individual will swing unless an extensive psycho analysis experiments and results for that individual are evaluated, but at a group level probabilistic behavioral outcomes are being done in many fields.

The point in question is AI, the latest and the greatest of the current buzz words has arrived. Soon it will be difficult to distinguish if you are talking to a man or machine online or on phone (except for the very balanced voice of Karen Jacobson telling you she is recalculating, and you almost feel she is in the car). These new developments have led some very ambitious projects and the latest ones I read about is to develop management decisions by experimentation, recording and then "algorithmising". This is quiet an interesting approach, and probably will work in certain situations. Hedge funds that can crunch 150 million signals into few decisions should embark on a more precarious route of mapping human interactions to simple outcomes. The problem is that those decisions are based on infinite other "nuances" of individuals that Bridgewater probably considers noise that can be suppressed. For one thing, the decision making by computer brains is easy. It is devoid of the emotional manipulations and political consequences and hence really, shall I say boring; easy nonetheless and in some cases where very defined paths must exist, efficient. Humans are also leaving an ever increasing trail of their activities and decisions following those activities, that make defining inputs and outputs of machines more accurately "human". The (people with)machines as it so happens will always carry a temporary advantage because in hyper competitive market nano seconds can switch millions. That being said, what one can "predict" using the same algorithms using similar data points very quickly becomes available to all market participants at some cost. The game ultimately returns to humans to weigh the inputs "calibrate" and then interpret the outputs, and the betting begins again.


The digitopoly blog talks about judgement becoming the currency of the day in the machine age, and what is funny about that is judgement is always the most precious of gifts at the higher echelons of management. The conundrum is judgement skills vs prediction skills; can one really predict without judgement which is a direct result of experience (practice and observation). Prediction in this sense will be basically one more input point to the judgement, the weight of prediction on the eventual judgement in this case will be more, simply because the machine will evaluate far more inputs in a shorter amount of time, more frequently, and arrive at a logical conclusion with consistency, accuracy and speed.

The improvements in AI are an extension of computing power. Computers are leveraged to do great many tasks but depend on humans. A certain type of knowledge is being created by the immensity of data and algorithms. Unique understandings from the cross section of this immense data and algorithms/models are forming the basis of new type of knowledge. This knowledge has its own basic blocks that can form bigger themes and simple conclusions around those themes under known set of conditions, inputs. More and more humans will master this knowledge and enhancements in AI will become extensions of humans just like language. In essence AI will not be AI but EI (enhanced intelligence) for humans.

Data few years later

We are back again with turmoil in thoughts riding on brain waves stimulated by all sorts of observations and intake of information, some voluntary other completely involuntary and when I say involuntary I mean as in predictably irrational sort of way, where I was messaged to be manipulated and I got tricked by my own programming.

So the question is if we are predictable in our behavior to certain degree and messages can be created to transform us or to "guide" us through our selection process, how will it all turn out for us as consumers of information, buyers of products, recommenders of services, spreaders of word, as the crafting, scaling and targeting of those messages becomes more granular, more scientific and shall we dare say more accurate.

Where will all this ability to forecast "us" lead to?

We say finding information has never been easier. Anybody who has gone through a book store or a library and tried to find a book through the ISBN numbers will tell you that Amazon, Google and the likes have "bettered" it. But if 90%+ people don't go beyond first page of search and search for most words results in millions of search engine result pages, is the problem really solved and why are all those other pages created, maybe that is question for another time but if more people "link" to information does that make it the most likely to be what I am looking, The answer is "most likely" yes!, but the prominence of sponsored results on all of these information platforms and users propensity to actually think they are relevant tell us that we are not as black and white, even though a tad bit predictable.

 Ah! but here is where we say that all that predictive goodness of machine learning will help since not all users are alike in their tastes and not all humans understand the intent of their words, the machine will guide them by looking at their past intents through their past journey and present them with the most likely result of their endeavor better than they intended. Currently scaling on the traditional online realm of internet, this type of "intelligence" will be soon intertwined with our lives. Never mind the privacy concerns, the thought of the cyber space and all kinds of devices knowing me and my intent better than I do myself is, well, I wouldn't say scary, it is just another phase of enlightenment.

Data, analytics, algorithms machine learning may seem like something from start trek to some, for now, but that will change. The users will casually plug in algorithms to find relevant information. No one will be concerned where the intelligence comes from Azure, Watson, Tensor Flow, Spark, Hadoop, Cloudera etc etc... I don't mean in the broader sense, because in broader mechanisms this is already in place. Companies and organizations already use this to find solutions, I mean at a user level.

 If in past I have felt comfortable at a certain temperature in my house, what 3 devices should be kept on and what should I eat to go with that temperature and my mood. Actually how should I be feeling in this ambient atmosphere, well maybe I will leave that for me to decide (but it sure can be influenced on a tiring day). If my watch is connected to my voice and body, the devices in my home are connected to my watch, and my voice is connected to my devices through "echoing" gizmo, and to my transport, it is all but a matter of crunching sensory information to predict what is going on with me at that moment and take or recommend the "most likely" actions.

I can be in my car, and all I have to say is "game day" or "super bowl party" and some of the things are taken care of automatically. The TV is set to turn on the time it needs to on the station it needs to, snacks (based on off course my "patterns") are already ordered, wine /beer ordered, the cleaning robot started, the cooking robot in action, the room temp adjusted, the door man or the Automata /Virtual receptionist informed....my intent of a get together posted....to media of choice....responses gathered ...and responded to... the list of consequential preparation can go on till all there is left for me is to get out of the car at my door, while the car goes and parks itself, pick up the stuff in my pickup spot (brought in by a flying droid...but of course) and listen to the status of the tasks..list of potential guests....weather..sports....and days proceedings... as I walk in the door....

I don't want to predict that HAL "3000" will be saying, sorry Ashar I can't do that, you must eat your no salt, non fat, no cholesterol..."Soylent green", but with all that understanding of behavior and its connection to future of a person, regulations will pop up, lets just hope that I can still have my fill of delectable indulgences.

The data is already ubiquitous and so is the collection. The interpretation and use is rapidly gaining ground in all facets of life. It will continue, and there will be a point where the discussion of data,  utilization will be moot. It will be a fact of life just like the internet.

Human Brain is not built to last in behavioral economics

I have started two classics at the same time by same authors. "Good To Great" and "Built To Last" and it just so happens that life and business cycles do happen and some companies go from good to great to dust just like us. That doesn't reduce the greatness of those books, not one bit. But it goes to show that like few good decisions on the tailwind of favorable economic and market conditions can propel company to great heights and keep it there for relatively long term if the good judgments continue, equally, few bad decisions combined with headwinds can wrap up a business and send its employees on permanent vacation. Lingering in my mind are the busy thoughts around them something fascinating, the human mind.

That is the human folly, the mind that makes us believe we made conscience decisions, in complete discernment. Amazing is that we are thrilled with psychological findings and their implications on our decision making, at the same time we know that biochemical reactions resulting from stimuli are causing those feelings, there are no feelings, but yet there they are tormenting us, exciting us, making us smile, frown, laugh cry. A training can generate certain emotions in one and none in another. So yes we are predictably irrational.  

The merriam webster dictionary defines heuristic as mechanism of learning and improving. The delicate part in the simple definition is the improvement. The mind does learn, but it learns pretty much as a computer. The stimuli to reaction is based on how the mental machine is programmed and programmable it is. Where the bits and bytes become the biases infused by circumstances and experiences and the verdict of words, actions, feelings, the output of those biases.

Few of the concepts that interest me:

1.    Association

2.    Anchoring

3     Priming

4.    Norms

5.    Halo effect

6.    Substitution

7.    Intuitive vs. Rigorous 

8.    Bias in Data

While the only thing I can gain from build to last and good to great is topics, reasons and actions to improve and acquire from research and conclusions that will be impossible for the normal of us to collect and evaluate on our own, there are times when larger than life seems an illusion presented as reality, a fantasy disguised as a fact and it is acceptable as long as it appears larger than life. In expanding universe, where million lights years long galaxies are formed, stars are born and stars die. I am looking at a dead star or a very bright star becomes irrelevant. The light from thousands of light years alludes to a shiny presence somewhere in time and space. Whether it is breathing fire at this very moment in time is too much decision making responsibility for my brain. Short of consulting Stephen Hawkins, I am good with believing that yes luck has a hold on everything in the great mirage, but the mirage is large enough for me to accept it as certainty. There are some universal truths that surround great companies and that companies are built to last, if they last longer than my life span.

The world of association

While Bayesian inference and statistical distributions are formidable topics and well worth the ink and mental exertion, when it comes to forecasting and customer modelling, my mind wanders from thoughts of aggregation versus innovation to queens gambit and then to thinking fast and slow. 

Half an hour of commentary on Petrosian vs. Kasparov and barely sweating on treadmill it is back to Kahneman. 

All businesses are based either on aggregation of goods and service or innovating a product or service. If a business is not innovating, it has to constantly aggregate to grow. Entering in to new geographies, trying to find new customers, expanding into new lines of goods, increasing share of wallet, acquisitions are all hallmark of collection based growth. Recently the nature of aggregation has changed. Things that were impossible to aggregate are now easily collected, shared,  regurgitated and with speed of dispersion that is extremely complex to measure or predict. Hard goods, soft thoughts and connections of all sorts, available for gathering. The innovation in gathering and packaging is an old art for businesses, but everything that was available before lacked scale due limitation of the medium is now easily created, destroyed, recreated and most importantly, connected. Scaling and aggregation of connections has opened up and disrupted established business norms. The change of medium is a paradigm shift.

Association analysis gives us the prediction, what goods have a higher chance of being bought with other goods. With basket analysis you can see the relation of product A with product B that only exists in the need of the buyer. There may be no other viable attribute relationship between the two products. Target' diapers and beer is probably the most famous example and next to it is Amazon's recommendation engine.

But products are predictable and easy to identify with availability of historical evidence. A lot more fascinating happening is the same concept in human mind. Priming, where words are associated in our mind, unbeknownst to us, with other words. The triggering of one word might act as a carrot to lure into mind the associated words. Even more curious is the discovery that word affiliations actually have physical implications.  Unfortunately or fortunately(I really am not so cozy with the idea of letting others into my minds thesaurus ) these associations are harder to measure. Nonetheless experiments have shown these associations exist and people tend to act predictably with mere stimuli of a word or an image. No wonder despite the obviousness of sales lures and ubquity of marketing slogans they still work.