Job extinction

For any young person about to start their career it's certainly worth thinking about what kinds of jobs are likely to be automated in the foreseeable future and basing your decisions around those expectations.  This article lists 20 jobs which may be in danger of extinction.

Some of these I definitely agree with.  Cashier, garbage collector, toll booth operator, news anchor, pilot and mail man all seem like jobs which if they're not fully automated in 20 years time will probably be significantly more automated than they are now, with fewer job opportunities.

In 20 years, or even in 50, I think there will still be human actors and actresses.  Yes, there are savings to be made with virtual actors, and yes the virtual actors will be highly realistic and glamorous, but I think there will be a backlash with audiences "getting tired of all these fake movie stars".  Some proportion of the acting profession will remain human, simply because people are inherently interested in other people and like to model themselves to some extent after their heroes - even if they're fictitious ones.  Also there is money to be made out of gossip and celebrity which would be hard to recreate with virtual actors.

I could turn out to be wrong about this, but I think that any occupation which involves accurate dexterous manipulation - especially of non-rigid objects - will remain expensive to automate in the foreseeable future.  Building machinery which recreates the movements and sensing of a human hand has not been easy to achieve, and even if it is achieved within this time frame it may remain too expensive for the mass market.  So I think jobs like car mechanic will still exist, although what car mechanics do may change since electrification means that the internals of a vehicle become radically simpler and the increase in use of electronics may mean that this job becomes more like that of an IT technician, doing non-traditional tasks such as sensor calibration.  In principle car maintenance could be fully automated, with appropriate design of the vehicle, but given the slow rate of change in the automotive industry I'm expecting only partial automation within two decades and perhaps more jobs created related to automotive sensing and safety systems and the checking/certification thereof.

Prostitution seems like an unlikely candidate for automation, but there would certainly be advantages to robot prostitution in terms of reduced issues with sexually transmitted diseases and zero probability of unwanted pregnancy.  However, similar to the issues with acting this is a human contact type of situation and I expect that robot prostitution will not be universally deemed to be the most desirable option.  Strange though it may seem, in the future people will still want to have sex with other people - at least some of the time.

Waiters will definitely see job competition from robots, and I expect that within 20 years it will not be unusual to be served by a robot in many dining locations.  However, I expect that human waiters will remain employed in the more upmarket restaurants as a sort of status symbol providing product differentiation from the riff-raff.  Also, jobs are likely to be created in the maintenance, programming and upgrading of robots performing customer-facing retail tasks.  A few jobs will also be created in the "back end" administration of teleoperator systems and teleoperator marketplaces, and of course a whole new market will open up for workers performing teleoperation tasks.

Librarian jobs will probably still remain, despite the near total electrification of books.  I think that libraries will turn into community centres where educational, training, voluntary and maybe even some medical activities take place, and that the job of the librarian will be more like that of a social worker or community manager.  Within the next two decades there will be a shake up of the higher education system, as traditional ways become less cost effective and the economic value of a college degree falls.  Libraries may be a beneficiary of this disruption as a classical university education becomes unaffordable for the majority.  This is really just a swinging back of the pendulum to a time prior to the baby boomer generation when libraries were the primary hub of popular education.

Assembly line workers have been under consistent threat of technological unemployment for the previous two centuries, and this trend is likely to continue.  What I've observed directly for myself is that the kinds of factory jobs which tend to be preserved and resistant to automation are those which require a high degree of dexterous manipulation - the "fiddly little stuff" - especially when dealing with objects which are non-rigid.

It almost seems silly to specify the job of "film processor", because this is a job which is largely already obsolete.  For the foreseeable future digital storage and distribution is the way that all media is going to go.

Security guards I think will be partially automated.  Jobs like patrolling and surveillance are on a trajectory to be fully automated within the foreseeable future, but those which involve human contact will remain either difficult or undesirable to automate.  Chasing then wrestling a suspect to the ground or bungling them into the back of a van is unlikely to be something which a robot could do easily, and even if this were possible there may remain legal liability reasons for having it done by a person.

Supermarkets and other big retailers have been trying to get rid of cashiers for quite some time, and it seems possible that within the foreseeable future they will eventually succeed both in installing appropriate technology and also persuading customers to use it.  I expect that a row of cashiers will be replaced by a couple of (human) security guards at the entrance and that paying for items will be done automatically via a mobile phone.  Probably for reasons of tradition as much as anything else shoppers will still place items onto some sort of conveyor belt where they will be counted and totalled automatically without a cashier.  Any item which cannot be counted (eg with a defective RFID) will be flagged, and the shopper directed to either replace it or put it into a reject bin.  An overhead camera above the belt, with cameras also elsewhere around the store, will deliver internet based cloudsourced security, with the security guards being alerted to any shoplifting or attempts to bypass the conveyors.  Cloudsourcers receive payments if the guard confirms that something suspicious occurred, or have reputation points deducted if it was a false alarm.  The lower your reputation, the less chance of being employed within the teleoperator marketplace, and equal opportunities employment regulations may mean that employers must always choose teleoperators with the highest ranking available at the time of recruitment - a process which can be fully automated.

Surgery will be more automated, but I expect that the automation in most cases will be merely augmenting the abilities of a human operator.  The article mentions malpractice cases, but it should also be noted that fully automated systems can make mistakes too, which may result in malpractice cases against the manufacturer - potentially a bigger problem than a malpractice case against an individual person.  Since surgery involves manipulation of flexible objects, within the next two decades I think that this is unlikely to be fully automated with a high degree of confidence that mistakes can be kept to an acceptable level, except in some niche areas such as eye surgery.  A career as a surgeon still seems reasonably safe.

Construction working also is a mixed bag in terms of employment.  There will be ways of constructing buildings which are much more automated than how things occur now, but to what extent these will be deployed remains unclear.  I'm not an expert on the construction industry, but expect that the cost of human labour is only a very small component of the overall cost of new building, with the biggest cost being merely buying land.  Whilst the capital costs significantly outweigh the labour costs there may not be very much motivation to introduce more automation.  New jobs in construction automation will be created in building, selling, maintaining and operating new types of robotic construction machinery.

Scare tactics

It looks as if Ben Goertzel is coming around to my way of thinking about SIAI, or perhaps he held this opinion all along but didn't say much about it.

"Personally, I'm a lot more worried about nasty humans taking early-stage AGIs and using them for massive destruction, than about speculative risks associated with little-understood events like hard takeoffs."

The problem with The Scary Idea is that it's just not elucidated in enough detail to be credible.  Certainly once you get into the realm of AI being embodied, either in large or small format, there are all sorts of hard physical constraints and practical issues which the doomsters consistently hand wave over.  They also tend to view intelligence as something which can be extended to indefinitely high levels, and not as a form of adaptation to an environment.  So in my view The Scary Idea is actually a bogus idea.  It's an idea that will lead you astray and waste your mental resources if you provide it with cognitive free reign.

I think all this scaremongering emerged particularly strongly in the years after 2001, when it was compellingly demonstrated that fear could be used to both mobilise public opinion and generate a revenue stream.  In the previous decade AI research seemed to have been languishing in one of its winter periods, but in what was commonly referred to as "the post-9/11 era" highly speculative, poorly characterised or astronomically minuscule risks were suddenly treated with a far greater degree of seriousness.  If AI could somehow be conflated with fears about public safety and terror plots then perhaps a new spring time of research funding would be forthcoming.  I see The Scary Idea as being one of the outgrowths of that zeitgeist.

In general it's probably not a good idea to have your thinking primarily dominated by fear.  Fear limits intellectual scope, encourages paternalism, and can result in overly conservative risk-averse decision making strategies along the lines of the precautionary principle which would appear to be at the heart of The Scary Idea.

Don Quixote complex

From an anthropological perspective there are also status implications for The Scary Idea.  Engineers, and especially AI researchers, are low status individuals in most societies.  Look how much they get paid/noticed compared to other professions or academic fields.  As far as I'm aware nobody has ever been awarded a Nobel prize in science for doing AI research.  Contrary to this lowly position The Scary Idea posits the AI researcher as a high status heroic character, striving to save humanity from A Terrible Fate.

For some previous opinions of SIAI and Friendly AI see Singularity Research Challenge: funding the wrong stuff and What is friendly?

New applications of coffee

This seems like a really good idea for a robotic gripper.



It would be cheap to manufacture and can handle a wide range of objects. A design of this kind radically simplifies both the mechanical and computational requirements for a "tidying up" kind of robotic application, eliminating the need for complex sensing of the shape of the object and motion planning of fingers. Even if a soft manipulator wears out or breaks after a few million cycles, it wouldn't be particularly expensive to replace.

Today's robot soldiers are tomorrow's robot police officers

Still largely a surveillance device, but moving towards the role of a soldier is this vehicle.  It's design doesn't look terribly well thought out.  The slope of the front of the vehicle no doubt provides the sensors with an excellent view of the road ahead, but it will also conveniently funnel dirt into what looks like a bank of sensors and also if someone shoots at the front of the vehicle then bullets are likely to deflect straight into the sensors, which might result in the vehicle being disabled.



The future of warfare on the ground is probably going to consist of a range of mostly teleoperated vehicles.  At the lower end you'll have the simple buggy or tracked platform with a gun strapped onto it, with something like a mobile phone at its heart.  These are essentially the disposable infantry of the future, and are the preferred modus operandi for terrorists or paramilitary organisations due to being cheap and easy to build in a hurry from commonly available materials.  At the upper end you'll have larger heavily armoured vehicles similar to the above, with a range of expensive sensors and having semi-autonomous navigation and perception for detecting other vehicles or people.

It's very likely that policing will follow the same trajectory, although lagging military development due to the many additional safety requirements.  Policing via UAVs is already beginning, but this isn't ideal for street level activities due to variations in weather conditions, obscuring trees or buildings and so on.  If the vehicle is capable of autonomously following a predefined patrol route then searching for suspicious activity can be crowdsourced via the internet, driving wages down to the lowest possible level and using labour from any geographical location.  Such a vehicle could be on patrol almost around the clock, stopping only to refuel or have its batteries changed, which provides the high availability/visibility which crime-ridden communities demand whilst also minimising labour cost, and could be armed with some non-lethal weapons such as sonic devices or water cannons.

In case anyone is worried about the accountability issues of say policing in England being carried out by teleoperators in Nigeria or elsewhere then a system of checks could be employed.  On the one hand you have the operators doing the policing, and on the other you have operators being paid to rate their actions as legitimate, according to some given rule set, or illegitimate.  If a teleopertor gains enough illegitimate points then they are excluded from the system, thereby losing their income (the equivalent of being fired in a teleoperated economy).  Also there would be the opportunity for communities to become self-policing by volunteering to do the teleoperation job on a part time basis.  If enough volunteers were available then the policing labour cost could fall to something close to zero.

Conventional policing would still be needed to do jobs which require the manhandling of suspects, but I expect that bundling people into the back of vans is a minor component of overall police expenditure and teleoperators could call in a dedicated (outsourced) squad to do this as necessary.

Living on a plane

On the effects of losing stereoscopic vision.



This is interesting because the conventional view about the function of biological stereo vision seems to be that it only applies at short distances of a few metres, or not that much more than arm's length, for things like grabbing the next branch or judging how to pick up an object.

Triage

There's a rather silly proposal to remove malware infected PCs from the internet apparently being made by Microsoft, accompanied by some laughably tame criticism of the security of Microsoft operating systems.

It all sounds rather dystopian, but somehow I think the chances of a system like this going ahead are incredibly slim indeed.  Apart from the general kerfuffle which would be caused, such a strategy would result in the biggest migration of PC users from Windows to Linux ever known.

Perhaps they're thinking that this would maximise revenue for the companies selling "security" software, and increase the amount of business coming from support contracts, as in the Cyber Clean Center.  There might be some truth in this, and when you take a look at the companies participating in the Cyber Clean Center, they're all the usual suspects who would stand to benefit financially from a more coercive approach to network security.

Escape from Alcatraz

My guess is that Facebook have done this preemptively to guard against the possibility of later legal trouble or anti-monopoly regulations.  Google did something similar with their Data Liberation Front. It also provides a clearer divide between users legitimately doing what they want with their own data and bots or screen scrapers which might be used for more nefarious purposes, such as by spam or phishing.



Unlike as suggested in this discussion of the Facebook movie I don't believe that there is any "major generational shift" with regard to privacy, in terms of some whole new attitude which will become the new norm.  I think it's just that if you're older you've accumulated more kipple in terms of personal history which potentially could have negative effects if it was all completely in the public domain.  Also older people are more likely to have health problems, and any revelations with regard to health can have quite substantial negative implications with regard to acquiring jobs, dating or other social activities.  For example, if a high ranking politician admits to any moderately interesting health problem there is an immediate media storm of speculation and the sharks begin circling waiting for any sign of weakness which can be exploited to their advantage.