The employment Implications of AI for the 'predictive' professions.

[cock-eyed octopus]

4 hours ago, scottbeard said:

That's just incorreect.  I am an Economics graduate.  Economists use mathematical techniques to help governments and business owners manage the allocation of scarce resources.  Of course the use of Maths doesn't mean that Economics is a natural science, and I have never heard any Economist pretend otherwise.  Indeed, human behaviour is ever-changing (unlike natural sciences) so many economic problems can never be completely "solved" the way that, say, E=mc^2 is a solved equation.

Your logic seems to be: Economists cannot predict anything with certainty, therefore they are of no more value than witch doctors.  Just as with weather-forecasting, or business planning, or government forecasting the point is that even though you don't know the future for CERTAIN, you still need to make decisions about it.  For example, a shop needs to buy an amount of stock in, even though it isn't CERTAIN how much their customers will want to buy next week.  You need to either take your umbrella with you OR not, even though you can't be CERTAIN whether it will rain.

Economics is more valuable than witch doctoring, because - even though it isn't perfect - it *is* better than no information at all when making your decisions.  And AI will no doubt help economists create even more helpful information - but whilst I can see AI helping with mathematical modelling it is harder to see it replacing the element of psychology.  And it will never be 100% accurate - how could any predictive process ever be?

Something better than 50% would be nice though. Once predictions fall below that level they are useless anyway.

Also, different economists frequently make completely different predictions - & none of them seem to be make consistently useful ones.

 

[scottbeard]

51 minutes ago, ****-eyed octopus said:

Also, different economists frequently make completely different predictions - & none of them seem to be make consistently useful ones.

 

Different doctors frequently make different diagnoses.  This is why people get second opinions on tricky cases.  However, saying "doctors are not always correct, and do not always agree - therefore doctors are of no value" is patently absurd, and yet that's the view you're taking about economists.

[cock-eyed octopus]

2 hours ago, scottbeard said:

Different doctors frequently make different diagnoses.  This is why people get second opinions on tricky cases.  However, saying "doctors are not always correct, and do not always agree - therefore doctors are of no value" is patently absurd, and yet that's the view you're taking about economists.

If doctors' diagnoses were as accurate as economists' predictions they would have no credibility. All experts stand or fall by the accuracy of their predictions (" this tower block will not burn"). As the accuracy approaches 50% their value approaches zero.

 

[zugzwang]

9 hours ago, scottbeard said:

That's just incorreect.  I am an Economics graduate.  Economists use mathematical techniques to help governments and business owners manage the allocation of scarce resources.  Of course the use of Maths doesn't mean that Economics is a natural science, and I have never heard any Economist pretend otherwise.  Indeed, human behaviour is ever-changing (unlike natural sciences) so many economic problems can never be completely "solved" the way that, say, E=mc^2 is a solved equation.

Your logic seems to be: Economists cannot predict anything with certainty, therefore they are of no more value than witch doctors.  Just as with weather-forecasting, or business planning, or government forecasting the point is that even though you don't know the future for CERTAIN, you still need to make decisions about it.  For example, a shop needs to buy an amount of stock in, even though it isn't CERTAIN how much their customers will want to buy next week.  You need to either take your umbrella with you OR not, even though you can't be CERTAIN whether it will rain.

Economics is more valuable than witch doctoring, because - even though it isn't perfect - it *is* better than no information at all when making your decisions.  And AI will no doubt help economists create even more helpful information - but whilst I can see AI helping with mathematical modelling it is harder to see it replacing the element of psychology.  And it will never be 100% accurate - how could any predictive process ever be?

There's little or nothing in existing micro- and macroeconomics texts (Samuelson, Mankiw, Varian etc.) that's worth reading. Economists simply haven't understood how to model markets mathematically in an empirically correct way. The reigning neoclassical theory based on assumptions of optimising behaviour and infinite foresight has been falsified. In fact, the physicist MFM Osborne offered a comprehensive refutation as long ago as 1977: The Stock Market and Finance from a Physicist's Viewpoint. Which isn't to say that economics can't be made into a science if it's done properly, see for instance the work of Fisher Black, Doyne Farmer or Steve Keen.

[scottbeard]

19 hours ago, ****-eyed octopus said:

If doctors' diagnoses were as accurate as economists' predictions they would have no credibility. All experts stand or fall by the accuracy of their predictions (" this tower block will not burn"). As the accuracy approaches 50% their value approaches zero.

"As the accuracy approaches 50% their value approaches zero" is a truism, but do you actually have any data that suggests that is the case?  In any event, Economists are not predicting "will burn / will not burn" events like a natural scientist.  If an economic forecast suggests that RPI inflation over the next 2 years will average around 3.0%, if the actual figure is 3.2% would you regard that as a success or a failure?

17 hours ago, zugzwang said:

There's little or nothing in existing micro- and macroeconomics texts (Samuelson, Mankiw, Varian etc.) that's worth reading. Economists simply haven't understood how to model markets mathematically in an empirically correct way. The reigning neoclassical theory based on assumptions of optimising behaviour and infinite foresight has been falsified. In fact, the physicist MFM Osborne offered a comprehensive refutation as long ago as 1977: The Stock Market and Finance from a Physicist's Viewpoint. Which isn't to say that economics can't be made into a science if it's done properly, see for instance the work of Fisher Black, Doyne Farmer or Steve Keen.

Unfortunately I have not read all of the texts concerned so I can't comment in detail on this, other than to note that Economists do an awful lot more than just estimate what will happen to the stock market.  It answers questions like: "the government wishes to spend £1 million to help disabled people in the North.  how should that be allocated between the North West and the North East to make best use of the money?" or "the government wishes to save £1 million from the council budget in Cornwall.  which services should be scaled back to cause the least damaging effects?"

[cock-eyed octopus]

3 hours ago, scottbeard said:

"As the accuracy approaches 50% their value approaches zero" is a truism, but do you actually have any data that suggests that is the case?  In any event, Economists are not predicting "will burn / will not burn" events like a natural scientist.  If an economic forecast suggests that RPI inflation over the next 2 years will average around 3.0%, if the actual figure is 3.2% would you regard that as a success or a failure?

 

 

If their prediction is no more accurate than that of The Man On The Clapham Omnibus I would regard it as a waste of everyone's time.

 

[wonderpup]

 

Quote

Your logic seems to be: Economists cannot predict anything with certainty, therefore they are of no more value than witch doctors. 

No- my argument is that Economists have the same value value as witchdoctors because they provide the same illusion of mastery and control over future events that in reality cannot actually be predicted or controlled.

The complex math calculations in which economists engage are best regarded as analogous to the rituals performed by the Withdoctor as he casts the runes in order to decipher the course of future events.

And just as in the case of the Witchdoctor, when the predictions of the Economist fail to materialise his status does not change, because his value to the tribe is not in the accuracy of his predictions but in the illusion of control over the unknown future that he represents. The reason Economists get hired in not because they can predict the future but because they offer reassurance to the employer's  clients that the future can to some degree be anticipated and thus controlled- this is a valuable contribution to the employer's bottom line.

So my argument is that the value added by those in the 'predicitive' professions-like economists- is not that they have any real ability to predicit anything but that they provide a comforting illusion of mastery and control over future events that allow their employers to reassure nervous clients that their money is in good and capable hands.

The reason that AI represents a threat to these professions is because it can viewed as an alternate means by which Clients can be reassured. If the meme that AI is a superior methodology for anticipating future outcomes becomes dominant the role of human experts like Economists will be diminshed as they find themselves either as appendages of or entirely replaced by Artifcial Intelligence systems.

It may not even be true that AI systems are any better at guessing the future than humans are- what matters is that this is believed to be true- at which point clients will demand that those managing their affairs employ AI systems rather than humans.

 

[stormymonday_2011]

1 hour ago, wonderpup said:

 

No- my argument is that Economists have the same value value as witchdoctors because they provide the same illusion of mastery and control over future events that in reality cannot actually be predicted or controlled.

The complex math calculations in which economists engage are best regarded as analogous to the rituals performed by the Withdoctor as he casts the runes in order to decipher the course of future events.

And just as in the case of the Witchdoctor, when the predictions of the Economist fail to materialise his status does not change, because his value to the tribe is not in the accuracy of his predictions but in the illusion of control over the unknown future that he represents. The reason Economists get hired in not because they can predict the future but because they offer reassurance to the employer's  clients that the future can to some degree be anticipated and thus controlled- this is a valuable contribution to the employer's bottom line.

So my argument is that the value added by those in the 'predicitive' professions-like economists- is not that they have any real ability to predicit anything but that they provide a comforting illusion of mastery and control over future events that allow their employers to reassure nervous clients that their money is in good and capable hands.

The reason that AI represents a threat to these professions is because it can viewed as an alternate means by which Clients can be reassured. If the meme that AI is a superior methodology for anticipating future outcomes becomes dominant the role of human experts like Economists will be diminshed as they find themselves either as appendages of or entirely replaced by Artifcial Intelligence systems.

It may not even be true that AI systems are any better at guessing the future than humans are- what matters is that this is believed to be true- at which point clients will demand that those managing their affairs employ AI systems rather than humans.

 

An interesting point.

One of the reasons people have faith in the predictive professions is that they believe they will somehow forewarn and protect them against future dangers. The problem is that while we know such events as calamitous floods, pandemics and economic crashes are historically inevitable the details of when and where they are going to happen is often inherently unpredictable. Even when there is significant amounts of data to show that such an event may be looming such as in the run up to the banking crisis no one seems to get the timing or location of the breaking point right.  Moreover since most people have a vested interest in the sky not falling on them human psychology tends to make individuals prefer to listen to those who predict it is not going to happen rather than the doom sayers who forecast the opposite. In that respect people want the predictive professions to protect them from calamity by simply not forecasting it. Given that calamitous events are rarer than mundane ones it is not surprising that the predictive professions have as tendency to play up the likelihood of the latter rather than the former particularly if that is what their clients want to hear. This has tended to lead to to the underestimation of fat tail risk in areas such as finance.

The question I suppose I would pose is that while AI might be better at forecasting such extreme risk events would human beings actually change their behaviour on the basis of such predictions particularly if it was something they did not want to hear? My suspicion is that they would not. In that respect the likelihood of AI replacing humans in the predictive professions might have less to do with how accurately they predict the future and more to do with the fact it may be a cheaper way of validating existing human tendency to underestimate true risks.

Edited June 29, 2017 by stormymonday_2011

[wonderpup]

Quote

The question I suppose I would pose is that while AI might be better at forecasting such extreme risk events would human beings actually change their behaviour on the basis of such predictions particularly if it was something they did not want to hear? My suspicion is that they would not. In that respect the likelihood of AI replacing humans in the predictive professions might have less to do with how accurately they predict the future and more to do with the fact it may be a cheaper way of validating existing human tendency to underestimate true risks.

I think you are right- it's the 'inconvenient truth' factor- we tend to ignore predictions we don't like.

What's interesting is that almost every human society in history has had it's oracles- they go by different names, like Shaman, or Witchdoctor or Economist- but their role is always the same, which is to act as mediators between the tribe and it's uncertain future. But our modern prognosticators claim that their oracular power comes from science, not superstition, pointing to the mathematical rigor of their craft.

But why would anyone who really could predict the future waste their time being an economist? Would they not simply use this power to make themselves rich?

Also it may in retrospect have been an error on the part of economists to insist that their oracular insight was the product of the rigorous application of math to the data, given that this is a domain in which computers may have a distinct advantage. It might have been wiser to emphasize the role of intuition and human psychology- but this was viewed as effete unmanly activity best left to the 'social' sciences. Economics wanted to be a 'hard' science, so much so that they even invented their own nobel prize to enhance their status in this regard.

In reality Economists are just another manifestation of the all too human desire to see into the future and thus control it, and in that sense an AI system designed to model future outcomes is also not a new idea- it's just yet another iteration of that same desire.

The attraction of a non human Oracle to the modern mind is paradoxical in that the very opacity of an artificial thinking machine would lend it a mystique that could transcend the established cult of the human 'expert'. We might find that the predictions of machines are held in higher regard than those of humans precisely because they originate from a non human source. Were that to happen we would have come full circle and ended up recreating in modern form the  oracles of the ancient past whose power of prophecy was derived from non human entities whose insight into the future was considered superior to that of mere human interlocutors.

If the reality is that those humans currently employed as oracles cannot really predicit the future with a usefull degree of precision- and this seems to be the case- then their job security would appear to rest on the their value as psychological support for the anxious client- however we already see AI developers bringing their own oracles to market, backed by the potent mystique of 'artificial intelligence' as a modern mythology that seems likely-in the long run- to eclipse the hitherto unchallenged dominance of human expertise.

 

[wonderpup]

Some straws in the wind here;

 

Quote

 

Credit Suisse’s head of data science and head of macro investing quit to set up their own AI firm

Two senior Credit Suisse staff have quit to set up a new company together.

Aric Whitewood, the bank’s head of data science, and Jonathan Wilmot, the head of macro investments in its asset management unit, have left to set up WilmotML, a company which promises to combine artificial intelligence and macro risk awareness to create advisory and intelligence products for a “low return world”.

 

http://news.efinancialcareers.com/uk-en/279189/wilmotml-aric-whitewood-jonathan-wilmot-credit-suisse/

This paragraph from the article is indicative of the kinds of narrative that will be constructed to support the idea of AI as the superior source of prediction;
 

Quote

 

A few years ago Wilmot told Bloomberg that the best advice he’d ever received was to keep your emotions in check when managing money. Maybe it’s better to get emotionless machines to manage money for you instead?

 

It's from such narrative speculations that meme's and mythologies are born.