Sharks Like Ice-Cream
29th of April, 2020
Fun Fact – Did you know if you eat ice-cream you are more likely to be eaten by a shark?
I am not a tech wizard as those who have spent any time around me know; I can make the strangest things happen technologically.
Some might say, but that doesn’t mean I don’t respect technology, what it can do and has done to improve our standard of living. However, I’d like to preface this blog with the “I’m no rocket or data scientist” disclaimer.
Machine learning is something that I find fascinating and is certainly living the hype at the moment.
There is just so much potential for this type of stuff …
Why are we so scared of Deep Learning Machines?
The idea of AI scares the hell out of most people. It is not hard for your imagination to run wild and end up thinking of the apocalyptic movies like The Terminator where the humans are enslaved and targeted for extinction by technology that escapes their control.
It’s a chilling thought to think it was only in 2029 that The Terminator was sent back from the future to hunt down Sarah Connor; that’s less than 10 years from now….
The idea of some black box performing human like feats or mimicking human ingenuity can be quite scary.
Elon Musk once said “I have exposure to the very cutting-edge AI and I think people should be really concerned about it…I keep sounding the alarm bell but until people see robots going down the street killing people, they don’t know how to react, because it seems so ethereal.”
Physicist Stephen Hawking said “the development of full artificial intelligence could spell the end of the human race.”
Associate professor Kilian Weinberger says that our negative feelings about AI can generally be divided into two categories: the idea that AI will become conscious and seek to destroy us, and the notion that immoral people will use AI for evil purposes.
AI might be proficient at tasks like quickly identifying images, patterns and even playing chess but there has been absolutely no progress in the area of AI attaining consciousness. So hopefully we will be safe from the likes of Hal from “2001: A Space Odyssey” for a bit longer.
As we know, humans are not always nice creatures and the idea of AI being used for harmful purposes is much more likely. Unfortunately like a knife, AI can be used for good or evil….
Correlation or Causation
But to me there is another issue with AI. One that creates massive opportunity in addition to the massive technological gains it will deliver.
And that is as these machines get smarter and smarter and developers and scientists continue to take for granted the things these machines are “learning” more and more, that it’s more likely that the new-found knowledge will be based on falsehoods.
“As humans we have been conditioned to look for causes, even though searching for causality is often difficult and may lead us down the wrong paths. In a big-data world, by contrast, we won’t have to be fixated on causality; instead we can discover patterns and correlations in the data that offer us novel and invaluable insights. The correlations may not tell us precisely why something is happening, but they alert us that it is happening. And in many situations, that’s good enough” Viktor Mayer-Schonberger & Kenneth Cukier, Big Data
WOW – that is a big statement! Go on read it again.
That is a big leap of faith we are taking. The movement away from understanding Causation to the acceptance of Correlation is a massive change in human approach to researching and learning.
Ice-cream and Sharks
I love fun facts! That’s why I started this blog with the fun fact:
Did you know if you eat ice-cream you are more likely to be eaten by a shark?
That is what the data says has been “Proved” many times!
Is this an example of Causation or Correlation?
The accepted response is that there is a “correlation” between those eaten by sharks and those eating ice-cream. This is brought about by the weather. The hotter it is, the more likely you are to consume ice-cream. You are also more likely to go to the beach and swim thus playing Russian roulette with our big white pointers…
Machine learning will find enormous amounts of correlations and causation in the massive data sets they crunch. It’s likely they will derive the most bizarre “facts” according to the data, upon which further learning may be based.
Rubbish in. Rubbish out.
As these models become more and more complex and the propeller head programmers become further removed from understanding what basis these machines are making their assessment and predictions on, the more errors they are likely to make.
AI is creating black-boxes that are becoming so complex that neither the programmers or anyone else really understand what is driving them.
Even worse is the fact that those building these machines have little interest in causation preferring to rely on the output from their Big Data sets.
Mike Flowers, the former Chief Analytics Officer for New York City famously once said “We have real problems to solve, I can’t dick around, frankly, thinking about other things like causation right now”
Get some data, feed it to the black box, give it a shake and let’s see what falls out….
Answer: “All cats have four legs, so all animals with four legs are cats”
Damn the logic.
Now I know this is a silly example, but I make it to show you the potential we have with this stuff.
How will this affect loan assessments?
Just think about it; the second half of this cycle is likely to see the rise of the Nano-Banks where loans and credit assessment will be created based upon the learning of these AI lunatics.
It’s hard to see how “Correlations” will not be accepted as “Fact” when assessing applicants.
And what data sets will be used?
We know that the finance industry always thinks that this time is different!
That lending behaviours from last cycle, let alone previous cycles, is irrelevant history!
So, the data sets, although they will be HUGE, are unlikely to contain long histories and will be based upon what the Machine Learning Spoofs build their algorithms on.
Garbage in. Garbage out.
We know that as a species, when not under pressure or filled with emotion, we can make reasonable decisions.
However, as soon as we are put under pressure or placed in a highly emotive situation god help their little Black Box assumptions….
It’s a space that is moving very quickly. Many will say I have luddite thinking as this time it is truly different. But isn’t it always?…
For me, it’s just something to be aware of as this cycle progresses. If we see the Nano Banks yelling, they have cracked the code, we will know we are not only in a bubble but heading for a crash.